Microsoft Word - numero_26_art_4


 

                                                                            S. Agnetti, Frattura ed Integrità Strutturale, 26 (2013) 31-40; DOI: 10.3221/IGF-ESIS.26.04 
 

31 
 

 

 
  
Strength on cut edge and ground edge glass beams with  
the failure analysis method 
 
 
Stefano Agnetti  
Department of Engineering, University of Perugia, Italy 
stefano.agnetti@gmail.com 
 

 
ABSTRACT. The aim of this work is the study of the effect of the finishing of the edge of glass when it has a 
structural function.  
Experimental investigations carried out for glass specimens are presented. Various series of annealed glass beam 
were tested, with cut edge and with ground edge. The glass specimens are tested in four-point bending 
performing flaw detection on the tested specimens after failure, in order to determine glass strength. As a result, 
bending strength values are obtained for each specimen.  
Determining some physical parameter as the depth of the flaw and the mirror radius of the fracture, after the 
failure of a glass element, it could be possible to calculate the failure strength of that. 
The experimental results were analyzed with the LEFM theory and the glass strength was analyzed with a 
statistical study using two-parameter Weibull distribution fitting quite well the failure stress data.  
The results obtained constitute a validation of the theoretical models and show the influence of the edge 
processing on the failure strength of the glass. Furthermore, series with different sizes were tested in order to 
evaluate the size effect. 
  
SOMMARIO. Il lavoro presentato ha lo scopo di valutare la resistenza di elementi in vetro. Il vetro, materiale 
elasto-fragile per la sua struttura chimica amorfa, raggiunge la rottura in modo improvviso per valori di 
resistenza piuttosto dispersi a causa della presenza dei difetti sui bordi del vetro.  
Sono stati eseguiti test a flessione su quattro punti su elementi in vetro float con differenti finiture superficiali, 
per valutare l’influenza della lavorazione dei bordi sulla resistenza. Sono state testate alcune serie di elementi con 
bordi tagliati e altre con bordi molati. Solitamente la rottura di un elemento in vetro avviene a partire del difetto 
di maggiore grandezza, nella zona maggiormente sollecitata. Grazie alla lavorazione dei bordi si riesce a ridurre 
l’ampiezza dei difetti, che, sebbene di minore grandezza, sono ugualmente presenti nel vetro. 
La resistenza del vetro è stata determinata attraverso la teoria della meccanica della frattura lineare e mediante 
l’analisi post-rottura della superficie da cui si genere la frattura. L’analisi sperimentale ha permesso di valutare 
l’efficacia della finitura dei bordi in termini d’incremento della resistenza del vetro e la validazione dei modelli 
per la determinazione della resistenza. 
 
KEYWORDS. Edge strength; Fracture mechanic; Glass flaw; Grinding; Size effect. 
 
 
 
INTRODUCTION 
 

or hundreds of years, glass has been used as windows in buildings, while research on structural applications of glass 
has only just begun. In recent years, the knowledge on glass properties has expanded, bringing to light new ways of 
using glass, such as full transparent construction. Due to transparent nature of glass, it is used in multiple ways by F 

http://dx.medra.org/10.3221/IGF-ESIS.26.04&auth=true
http://www.gruppofrattura.it


 

S. Agnetti, Frattura ed Integrità Strutturale, 26 (2013) 31-40; DOI: 10.3221/IGF-ESIS.26.04                                                                              
 

32 
 

engineers and architects and today the research is further advancing at the field of studying glass properties both in 
structural aspects and in relation to building technology, energy and light. 
But glass is a challenging material due to its brittle feature. In order to use glass safely in structural applications, knowledge 
about its strength is required. The presence of the flaw in glass causes the failure [1]. The fracture mechanics shows how 
the failure depends on the depth of the flaw, on the number of them and also on the stress corrosion (called static fatigue 
in literature) [2]. The stress corrosion causes subcritical crack growth in glass. The crack propagation phenomenon occurs 
in glass when it is exposed to tensile stress and humidity. The particular flaw that produces the fracture is generally called 
the critical flaw. 
The processing of the edge of structural glass is studied; the edge has an important role to determine the failure. Indeed 
the finishing of the edge could remove in part the flaws or in other case it could produce other micro-cracks, without no-
benefit for the strength of the glass [3]. The most important type of edge processings object of study [4] are grinding and 
polishing  
The strength of the glass can be evaluated through the fracture surface analysis: determining some physical parameter as 
the depth of the flaw and the mirror radius of the fracture after the failure of a glass element, it could be possible to 
calculate the failure strength of that [5]. For this evaluation, it was tested a group of glass element, in bending. It results 
that the edge processing has an influence on the failure strength of the glass. 
 
 
FRACTURE MECHANICS THEORY 
 

lass is an elastic material with a brittle behaviour at failure. Therefore linear elastic fracture mechanics (LEFM) is 
an ideal theory to model its behaviour. In fact, glass was the material used for the development of the basis of 
LEFM. In LEFM, mechanical material behaviour is modeled by looking at cracks. 

If we think at glass, as a material without flaws and defects, its resistance would be very high. But it doesn’t occur in 
practice because of the presence of the flaws. This phenomenon is explained by LEFM theory. 
According to the stress analysis conducted of an elliptical cavity in a uniformly stressed plate, the local stresses about a 
sharp notch or corner could raise to a level several times higher than the applied stress. It thus became apparent that even 
submicroscopic flaws might be potential sources of weakness in solids. 
Introducing the concept of the stress intensity factor (SIF), expressed to evaluate the failure, glass element fails when this 
value reaches the critical value KIc. 
The general relationship between the stress intensity factor KI, the nominal tensile stress normal to the crack’s plane σn, a 
correction factor Y, and some representative geometric parameter a, in general the crack depth or half of the crack length, 
is given by: 
 

 K Y a
I n

              (1) 
 

The fracture toughness KIc, also known as the critical stress intensity factor, is the SIF that leads to instantaneous failure. 
KIc is a constant value and is also called fracture toughness. Values KIc are available in literature, for soda-lime glass it is 0.75 
MPa m1/2. From LEFM is possible obtaining the failure stress form the measure of the depth of the flaw, as shown in [6] 
and [7]. 
 

Stress corrosion 
Glass is noted for its chemical inertness and general resistance to corrosion; therefore, it is used in the chemical industry 
and in the laboratory when chemical inertness is required. Despite this well-known property, glass is extremely susceptible 
to stress corrosion cracking caused by water in the environment. This phenomenon is known in the glass literature as 
static fatigue or delayed failure.  The susceptibility of glass to stress corrosion cracking was observed noting a time delay to 
failure and a loading rate dependence of strength. This effect is an activated process caused by water in the environment. 
Static fatigue of glass results from the growth of small cracks in the surface of glass under the combined influence of 
water vapor and applied load. 
Actually, glass is time dependent if it is in presence of humidity (only in vacuum it is time-independent). Stress corrosion 
causes flaws to grow slowly when they are exposed to a positive crack opening stress. A glass element stressed below its 
momentary strength (e.g. the static load) will still fail after the time necessary for the most critical flaw to grow to its 
critical size at a particular stress level. 

G 

http://dx.medra.org/10.3221/IGF-ESIS.26.04&auth=true
http://www.gruppofrattura.it


 

                                                                            S. Agnetti, Frattura ed Integrità Strutturale, 26 (2013) 31-40; DOI: 10.3221/IGF-ESIS.26.04 
 

33 
 

A kinematic relationship between crack velocity v and stress intensity factor KI exists and it is commonly used for glass 
lifetime prediction. For values of KI close to the critical value of KIc (that represents the glass toughness), v is independent 
of the environment and the crack  propagates very rapidly (for soda lime silica glass is about 1500 m/s). 
In the v-KI logarithmic-curve, v0 represents the position and n is its slope. Below certain threshold stress intensity Kth no 
crack growth occurs. The value of v0 and n parameters are discussed by Haldimann [1], for laboratory condition, v0 can be 
assumed equal to 0.01 mm/s, instead it is 6 mm/s in environmental condition. The parameter n is assumed 16. 
According to the theory of fracture mechanics, glass failure stress was defined using stress intensity factor KI. This 
equation is only valid in testing conditions where stress corrosion could be eliminated. If the subcritical crack growth is 
considered, the crack propagates as a function of loading time. This approach is presented by Haldimann [1], in which the 
crack velocity parameter v0 has the dimension of a velocity, instead n is dimensionless. 
 

  
2

2

0
0

2
( ) ( )

n
n ntn

nn
i Ic

n
a t a v K Y d

n
   


 

   
 

        (2) 
 

The same relation can be expressed in terms of the value σ with a static loading time. It is possible to obtain the failure 
stress σf, knowing the tf (s), i.e. the failure loading time and assuming aci (m) corresponding to the initial crack flaw depth. 

 

 

1

2

2
0

2

2

( )

( ) /

n

f f n
n

f Ic ci

t

t n v Y K a






 
 

  
  
 

       (3) 

 

Until a certain loading time, the inert strength is considered to determine the failure. But with the increase of the loading 
time, in presence of the stress corrosion, the relation (3) is used to determine the failure strength. The theoretical 
transition time loading between inert condition and time-depending condition, tref, that could be considered a reference 
value is obtained by Eq. (1) and (3). 
 

 
0

2

2( )
ref

a
t

n v



           (4) 

 

It t>tref the strength decreases following (3), instead if t<tref the failure is assumed to follow the inert strength level. 
 
Fracture surface analysis 
Fractography can bring quantitative information about loading condition at failure. The fracture surface is a source of 
information to determine the failure condition. Fractography of brittle materials is used to determine the origin of failure 
during strength testing, as in [8] and [9]. In general, this origin can be traced to material inhomogeneities, such as pores 
and micro-cracks, which occur due to machining (surface defects). Fracture features, such as mirror, mist and hackle 
zones, and crack branching, are formed upon failure. 
The fracture surface is a mirror zone that forms around the critical flaw, at the cross-section of the failed specimen. Under 
a failure stress, once the critical flaw starts to propagate, mirror boundary hackle lines are created after radiating crack 
reaches terminal velocity. 
The failure stress σf, i.e. the maximum principal tensile stress at the fracture origin, was approximately proportional to the 
reciprocal of the square root of the mirror radius (radius of the mirror/ mist boundary) rm: 
 

 1 2/f mr B             (5) 
 

Where B is a constant value (MPa m1/2), that depends on the material properties. However, limited informations are 
available about the time-dependency of glass strength in relation to the mirror radius. The relation (5) related to brittle 
materials, is valid for inert strength values. The time-dependency in glass strength is not taken in consideration in the 
measurement of the mirror radius. 
The analysis and interpretation of fracture mirror sizes in brittle materials are given in [5]. Fracture mirrors are revealing 
fractographic markings that surround a fracture origin in brittle materials. The fracture mirror size may be used with 
known fracture mirror constants to estimate the stress in a fractured component. Alternatively, the fracture mirror size 
may be used in conjunction with known stresses in test specimens to calculate fracture mirror constants. 

http://dx.medra.org/10.3221/IGF-ESIS.26.04&auth=true
http://www.gruppofrattura.it


 

S. Agnetti, Frattura ed Integrità Strutturale, 26 (2013) 31-40; DOI: 10.3221/IGF-ESIS.26.04                                                                              
 

34 
 

The relationship (6), called the crack branching equation, is generally accepted but its application is not always easy 
because reading the fracture pattern becomes quite complex for high stresses values that produce the failure. 
The crack branching equation is useful to evaluate the failure stress from the measure both of the mirror radius rm, both 
of the hackle radius rh, both of the half branching length rb (The subscripts m, h, b, represent respectively the mirror, the 
hackle and the branch): 
 

 1 2/f ar mr  
            (6) 

 

where σar is interpreted as being an apparent residual compressive surface stress and α is a constant value. Finally, all three 
branching constants αm, αh and αb as well as the corresponding apparent residual stresses σar were determined in recent 
studies, [10], and presented in [2] and summarized in Tab. 1. 

 

α [MPa m1/2] σar [MPa] 

Mirror 1.98 9.6 

Hackle 2.11 9.1 

Branch 2.18 10.7 
 

Table 1: Parameters of the crack branching equation for annealed glass. 
 

Size effect 
The size of the loading area has an influence on the failure stress [11]. This effect can be explained with the Weibull 
theory referring to the fact that a larger panel is more likely to have a large flaw in a high stress region, than small panel 
[12].  The relation that explains the size effect is: 
 

1

21

2 1

.

.

eff

eff

V

V




 
   
   

          (7) 

 

where σ1 and σ2 are tensile stress value for two elements of different sizes, of effective volumes Veff.1 and Veff.2. The β 
parameter can be calculated according to Weibull fit in EN 12603 [13]. 
In a recent study [14], it was tested the influence of the size effect testing a large number of samples. It is notice that larger 
elements tend to have a greater probability of larger flaws. The results demonstrate some differences between 
experimental and theoretical rates of the characteristic strength values, so, it means that the size effect has to be 
investigated more thoroughly. 
 
 
NUMERICAL RESULTS 
 

ive sets of elements were tested in four points bending test. In Tab. 2 the geometrical dimensions, the edge 
finishing and the number of the samples are shown. Sets A and B present the same geometry, but different edge 
finishing. Sets C and D have another size, in which the length is bigger than that of the former series. Finally, the 

size of the samples of set E is the biggest. Sets A, C and E present ground edges; instead sets B and D consist of glass 
beam with only cut edge. The ratio between height, h, and length l, of the elements of the various series is almost constant. 
The ratio between the load span, sl, and the support span, ss, are almost constant too. 

 

 
b 

[mm] 
h 

[mm] 
l 

[mm] 
Edge

finishing
N° of

samples
ss [mm] 

ls
[mm]

h/l ls/ss 

SET A 8 50 400 Ground 25 360 120 49.56 0.33 
SET B 8 50 400 Cut 25 360 120 35.01 0.33 
SET C 8 50 550 Ground 19 500 160 47.41 0.32 
SET D 8 50 550 Cut 18 500 160 109.7 0.32 
SET E 8 100 1100 Ground 20 1000 300 56.94 0.30 

 

Table 2: Geometrical dimensions, edge finishing and number of the samples for the various sets of glass elements. 

F 

http://dx.medra.org/10.3221/IGF-ESIS.26.04&auth=true
http://www.gruppofrattura.it


 

                                                                            S. Agnetti, Frattura ed Integrità Strutturale, 26 (2013) 31-40; DOI: 10.3221/IGF-ESIS.26.04 
 

35 
 

The specimens were subjected to in-plane four-point bending tests in a universal UTS testing machine (Fig. 1-a). The 
specimens were loaded at a stress rate of 0.75 MPa/s ± 0.15 MPa/s. For each sample the failure load Pf, the time at failure 
tf and the maximum displacement in the midspan wf [mm] was collected. The test ends when the beams collapse. The 
failure stress values or tensile strength value f, were calculated with the following equation: 
 

 
2 2

32

6

f

f

P
d

P d
f

h bh
b

 
 
             (8) 

 

where d [mm] is the distance between the load and the support; b [mm] is the width of the specimen and h [mm] is the 
height of the specimen. The failure initiated at the loaded area linked to the tensile stresses at the edge, creating a crack 
opening. The test set-up used is illustrated in Fig. 1-b).  
 

  
a) b) 

 

Figure 1: Test set-up used for in-plane four point bending tests (for each sample). 
 

The mean results for each set are shown in Tab. 3. As first result of the bending tests, it is possible to observe that the 
mean failure load for series A, having ground edges is higher than the one of series B. Instead comparing series C and D, 
the mean failure load for the set of beams with cut edge is slightly higher than the one with ground edge. The failure 
happened always in the zone inside the load span: cracks propagate from the tensile zone, at the bottom edge, to the 
compressive zone, in which the cracks branch out into several branches. The moment at failure was determined by the 
failure load.  

 

 
Pf [N] tf [s] Mf [Nmm] f [Mpa] wf [mm]

SET A 2290.14 75.54 148859.10 44.24 2.52 

SET B 1864.60 66.56 121199.26 36.43 2.22 

SET C 1624.27 50.10 138063.26 40.81 2.09 

SET D 1925.89 52.77 163700.74 45.65 2.76 

SET E 3446.28 59.88 603098.64 40.82 2.79 
 

Table 3: Mean values at failure for each series. 
 
Observing the various failures, it was possible to identify some particular crack patterns that are frequently presented (Fig. 
2). The crack pattern depends on the strength of the beams and directly it depends on the flaw measure. Beams having 
bigger strength (the flaws are small) presents more glass fragments, instead beams that collapse in a few number of pieces 
presents big flaws. 
The failure happened always in the zone inside the load span: cracks propagate from the tensile zone, at the bottom edge, 
to the compressive zone, in which the cracks branch out into several branches. 

http://dx.medra.org/10.3221/IGF-ESIS.26.04&auth=true
http://www.gruppofrattura.it


 

S. Agnetti, Frattura ed Integrità Strutturale, 26 (2013) 31-40; DOI: 10.3221/IGF-ESIS.26.04                                                                              
 

36 
 

 

  
Figure 2: Crack patterns of glass beams. 

 
As regard the test performed, the experimental results were analyzed with the LEFM theory and the glass strength was 
analyzed with a statistical study using two-parameter Weibull distribution fitting quite well the failure stress data. 
 
LEFM analysis 
Glass specimens were examined using an optical microscope with polarized light after the failure. Important information 
on the flaw sizes, the fracture origin and the fracture mirror sizes were collected. In the following Fig. 3-b some examples 
of fracture surface are presented. It is easy to identify the flaw and the mirror zone (shown in Fig. 3-a). 
 

 a) 
 

 b) 
 

Figure 3: Illustration of the depth flaw and of the mirror radius. 
 
The results of the post-fracture analysis are summarized in Tab. 4, where the mean values of the mirror radius and of the 
flaw depth are presented for each series. It is possible to observe that the ratio between the mirror radius, rm, and the flaw 
depth, a, is almost constant and it is about 9÷10. 
The strength σf(tf) is obtained by Eq. (3) knowing the measure of the flaw depth, a, the time at failure, tf and the critical 
value of KIc. The value of v0 and n parameters are discussed in [1]; for laboratory condition, v0 can be assumed equal to 0.01 
mm/s; the parameter n is assumed 16. The geometry factor, Y, is defined as a constant value for various edge crack 

http://dx.medra.org/10.3221/IGF-ESIS.26.04&auth=true
http://www.gruppofrattura.it


 

                                                                            S. Agnetti, Frattura ed Integrità Strutturale, 26 (2013) 31-40; DOI: 10.3221/IGF-ESIS.26.04 
 

37 
 

configurations; it is chosen by literature, [1], observing each crack pattern. In Tab. 4 the mean values of the strength, σf(tf), 
the time reference, tref, and the inert strength, σf(tref),  are shown, too. 
 

rm[mm] a[μm] rm/a σf(tf) [MPa] tref [s] σf(tref) [MPa] 

SET A 2.50 292.4 9.0 52.91 4.18 54.03 

SET B 3.89 456.1 8.4 45.43 6.52 46.50 

SET C 2.28 352.5 10.1 50.74 5.04 50.53 

SET D 2.31 247.1 9.4 50.14 3.53 50.21 

SET E 2.56 271.5 10.0 55.01 3.88 54.91 
 

Table 4: Mean values of the failure parameters, failure strength, time reference and inert strength for each series. 
 
Glass strength was analyzed with a statistical study using two-parameter Weibull distribution fitting the failure stress data 
as presented in [15]. The two parameter Weibull function is expressed as: 
 

 1( ) exp
x

G x




      
   

         (9) 

 
where: 
G(x) means a distribution function of x percentage of failure; 
θ is a scale parameter in Weibull two-parameter distribution; 
β is respectively a shape parameter. 
Weibull two-parameter probability distribution permits a correct analysis of glass strength. In Tab. 5 the two parameter of 
the Weibull distribution are presented. 
 

θ β 

SET A 4.161 59.555 

SET B 9.241 48.014 

SET C 4.694 55.128 

SET D 6.639 54.182 

SET E 3.166 63.726 
 

Table 5: Parameter θ and β of the Weibull two-parameter distribution. 
 
The diagram in Fig. 4 shows the fitting of the Weibull distribution for the series A, under the hypothesis that the 
specimens contain a random flaw population. The same diagrams were obtained for the other set of beams. The stress 
values vary largely in the range of 16.24 MPa to 97.35 MPa. The mean stress values vary in a range of 45.43 MPa to 55.01 
MPa.  The variation was noticed to exist between edge finishing; values of standard deviation vary from 5.5 to 16.4 for the 
five series of specimens. 
In order to compare the LEFM theory with the failure surface analysis, the parameters for the crack branching equation, α 
and σar, were determined by a linear regression in the diagram of the failure stresses and the mirror length. The mirror 
length was measured together with the flaw depth, in the post failure microscopic analysis. 
The measured length of the mirror was used to evaluate the crack branching Eq. and to establish a value for the branching 
constant. The objective is to compare the constants obtained from this analysis, using a trend line, with the ones proposed 
by the previous studies (Fig. 5). 

http://dx.medra.org/10.3221/IGF-ESIS.26.04&auth=true
http://www.gruppofrattura.it


 

S. Agnetti, Frattura ed Integrità Strutturale, 26 (2013) 31-40; DOI: 10.3221/IGF-ESIS.26.04                                                                              
 

38 
 

 

    
 

Figure 4: Weibull two-parameter probability distribution for sets A and B. 
 

    
 

Figure 5: Diagram of the measure of the mirror length rm (m-1/2) vs. the failure strength σf(tf) for sets A and B. 
 
The mean values of the parameters α and σar, calculated using the crack branching Eq., are 2.00 and 11.06. These are 
slightly higher than the ones proposed in Tab. 1 for the evaluation of the failure stress from the measure of the mirror 
radius rm, but they are in accordance with the ones proposed by literature. 
However, in general the inert strength is higher than the strength at failure time, when stress corrosion is considered. 
Values of the strength at failure time, σf(tf), and values of the inert strength, σf(tref),  are obtained and compared in the 
previous Tab. 4.  
As explained by the theory, inert strength is higher than the strength measured at failure, σf(tf) (Tab. 4). Stress corrosion 
law was obtained by the mean values for each series of beams and it was plotted together with the mean strength for each 
lot of samples in the following Fig. 6. 
 

 
 

Figure 6: Stress corrosion law for each series of samples. 
 
Size effect was studied considering sets A and C, having ground edges. Larger bodies generate lower mechanical strength 
values taking into account the higher probability of finding natural heterogeneities (flaws or cracks). In Weibull weakest 
link theory, the ratio between the mean mechanical strength values and the ratio of the effective volumes of the specimens 
leads to a strength dependency on body size as explained in the relation (7). For the analysis on the size effect for the 

http://dx.medra.org/10.3221/IGF-ESIS.26.04&auth=true
http://www.gruppofrattura.it


 

                                                                            S. Agnetti, Frattura ed Integrità Strutturale, 26 (2013) 31-40; DOI: 10.3221/IGF-ESIS.26.04 
 

39 
 

ground edge samples, the results in terms of failure strength show that set E present the greater values, that is higher than 
the failure strength of sets A and C; for this reason set E was excluded from the analysis. Instead the value of σf(tf) for the 
set A is higher than the one of the set C. In the latter case shorter samples present higher stresses that are what we expect 
by the theory: only sets A and B were compared. 
The analysis permits to obtain the same value 1.05 form the ratio of the strength σA/ σC that is equal to the corresponding 
(Veff,C/ Veff,A)1/β. 
 
 
CONCLUSIONS 
 

law characterization on glass beams was studied.  The experiments performed, confirm the applicability of the 
failure prediction theories commonly applied to explain the failure strength of glass.  
Each specimen was assumed to fail in mode I (in LEFM theory, mode I is a normal-opening, while modes II and 

III are shear sliding modes); the failures initiated inside the loaded area. The analysis and the measurements executed 
confirm the relation existing between flaw size and strength: the larger the critical flaw initiating the failure, the lower the 
strength. An equivalent relation is valid for the mirror radius: the larger the mirror radius, the lower the strength.  
The edge finishing produces advantage in terms of strength. The strength of ground edge glass is higher than the one of 
simply cut edge glass. The edge processing produces a reduction of the maximum depth of the flaw and an increase of the 
strength. 
Furthermore it was observed that the failure strength had no linear relation to flaw depth. In analogy, the fracture surface 
analysis shows that the failure strength had no linear relation to the mirror radius. 
In the LEFM the values of the geometric parameter Y, were estimated by observing the fracture surface, in order to 
obtain the failure strength for each sample.  In the fracture surface analysis, constant values α and σar, were confirmed 
using a linear regression law, fitting the measured data. 
The observations on the size effect show that the bigger is the length of the beam, the bigger is the probability to find 
flaws. In general, beams with edge processing, as grinding or polishing is recommended. 
 
 
REFERENCES 
 
[1] Haldimann, M., Fracture strength of structural glass elements – analytical and numerical modeling, testing and design, 

PhD These n. 3671, EPFL, Lausanne, Switzerland (2006). 
[2] Haldimann, M., Luibe, A., Overend, M., Structural use of glass, Structural Engineering Documents, IABSE - AIPC – 

IVBH 10 (2010). 
[3] Lindqvist, M., Vandebroek, M., Louter, C., Belis, J., Influence of edge flaws on failure strength of glass, In: 

Proceedings of GPD 2011, Tampere, Finland, (2011) 126-129. 
[4] Vandebroek, M., Belis, J., Louter, C., Van Tendeloo, G., Experimental validation of edge strength model for glass 

with polished and cut edge finishing, Engineering Fracture Mechanics, 96 (2012) 480-489.  
[5] ASTM C1678-10, Standard practice for fractographic analysis of fracture mirror sizes in ceramics and glasses, 

America Society for Testing materials (2010). 
[6] Carpinteri, A., Meccanica dei materiali e della frattura, Pitagora Editrice, Bologna, (1992). 
[7] Lawn, B.R., Fracture of brittle solids. 2nd ed. Cambridge: Cambridge University Press, (1993).  
[8] Dielhof, M. H., Dortmans, L. J. M. G. , de With, G., Fractography of Borosilicate Tested in Three- and Four-Point 

Bending Glass, Journal of the European Ceramic Society, 12 (1993): 215 220 
[9] Conoway, J. C., Mecholsky, J. J., Use of crack branching data for measuring near-surface residual stresses in tempered 

glass, Journal of the American Ceramic Society, 72 (9) (1989) 1584–1587. 
[10] Zaccaria, M., Overend, M., Validation of a simple relationship between the fracture pattern and the fracture stress of 

glass, In: Engineered Transparency. International Conference at Glasstec, Düsseldorf, Germany, (2012). 
[11] Fischer, H., Rentzsch, W., Marx, R., A modified size effect model for brittle non-metallic materials. Engineering 

Fracture Mechanics, 69 (2002) 781-791. 
[12] Quinn, G. D., Weibull Effective Volumes and Surfaces for Cylindrical Rods Loaded in Flexure, Journal of the 

American Ceramic Society, 86 (3) (2003) 475–479. 
[13] EN 12603:2002, Glass in building – Procedures for goodness of fit and confidence intervals for Weibull distributed 

glass stength data, CEN (2002). 

F 

http://dx.medra.org/10.3221/IGF-ESIS.26.04&auth=true
http://www.gruppofrattura.it


 

S. Agnetti, Frattura ed Integrità Strutturale, 26 (2013) 31-40; DOI: 10.3221/IGF-ESIS.26.04                                                                              
 

40 
 

[14] Vandebroek, M., Lindqvist, M., Louter, C., Belis, J., Edge strength of cut and polished glass beams, In: Proceedings of 
GPD 2011, Tampere, Finland, (2011) 476-479. 

[15] Weibull, W., A statistical theory of the strength of materials, Royal Swedish Institute for Engineering Research, 151 
(1939). 

 
 

http://dx.medra.org/10.3221/IGF-ESIS.26.04&auth=true
http://www.gruppofrattura.it