Construction Materials Proceedings of the Institution of Civil Engineers

Construction Materials 166 June 2013 Issue CM3

Volume 166 Issue CM3 Pages 158–174 http://dx.doi.org/10.1680/coma.11.00050
Undercut anchorage in dimension stone Paper 1100050
Received 17/08/2011 Accepted 17/10/2011
cladding Published online 02/08/2012
Camposinhos Keywords: anchors & anchorages/design methods & aids/
stress analysis

ice | proceedings ICE Publishing: All rights reserved

Undercut anchorage in
dimension stone cladding
Rui S. Camposinhos PhD
Coordinator Professor, ISEP, Polytechnic of Porto, School of Engineering,
Porto, Portugal

This paper analyses undercut anchorage technology, in particular its behaviour and performance as a fixing system for
dimension stone cladding of rainscreen façades. Based on a number of mechanical tests using various types of
Portuguese stone – three igneous, two sedimentary and one metamorphic – a study was carried out to investigate the
relationship between the flexural strength and breaking load of specific and very common stone types. Several
physical and mechanical characterisation tests and 130 pull-out tests with 6 and 8 mm cone bolt threads were
performed to determine the pull-out load failure on six different stone types: three granite, two limestone and one
marble. Finite-element stress analyses were carried out, and the test results were the basis for calibrating a simple
formula that can be used to estimate the stone’s breaking load at the undercut anchorage. Stress concentration
factors are proposed to take into account the undercut drill hole geometry and the specific properties of each type of
stone. Stone specimens from the same batches were subject to pull-out force tests using dowel anchorages, whose
values were then compared with the breaking load of undercut anchorages. Results are discussed and conclusions are
drawn based on tensile stress values by comparing the test results, the finite-element method and the proposed semi-
empirical formulations for the same breaking load.

Notation systems, using the more expensive fixing option, in terms of

Acf area of the projected cone failure surface materials and fixing, can offer other benefits resulting in cost
du anchorage depth and time savings in the overall project.
em minimum edge distance
F applied force at failure Resistance to lateral loading, mainly wind and seismic action,
Ku stress concentration factor is usually achieved by means of stainless steel anchors inserted
r radius of the specimen (disc) into kerfs or holes that are drilled or cut into the edges of the
sm minimum space between anchors stone panels. These anchors are mechanically connected to the
t thickness of the disc building’s structure, thus providing the essential mechanical
a average angle of the cone failure surface connection between the stone and the structure. One structural
am average spall angle weak point in this type of stone construction is to be found at
wt diameter of the cylindrical drill hole the kerfs or anchor holes on the edges of the stone slabs. These
wu diameter of the undercut cuts need to leave sufficient stone thickness to provide the
smax actual maximum stress necessary strength to resist the various forces or actions to
st tensile strength of the stone which the stone panels are subject.
stm medium tensile stress
sRf strength of a brittle material under flexure The failure of any stone panel may have significant con-
sRtf axial tension sequences. Many engineers and architects have an over-
sRf1 flexural strength of specimen with volume v1 confident attitude toward these issues and, although they
sRf2 flexural strength of specimen with volume v2 may not fully understand the engineering aspects of natural
stone and the behaviour of natural stone veneers, this
drawback does not seem to be a constraint when they specify
the selection, manufacture and use of natural stone as a façade
1. Introduction cladding material. This attitude, when combined with a lack of
Art and science must come together in the design of stone regulatory controls and standards stipulating appropriate
façades. The beauty and quality of natural stone make it an practices, has often led to situations of serious flaws in the
ideal material for prestigious projects using successful selection, design, and installation of natural stone cladding
construction methods in many countries. Rainscreen cladding panels.

158

Proceedings of the Institution of Civil Engineers - Construction Materials 2013.166:158-174.

 Construction Materials Undercut anchorage in
Volume 166 Issue CM3 dimension stone cladding
Camposinhos

Developing innovative fastening products requires in-depth stainless steel. The sleeve is made of stainless steel or carbon.
knowledge about the physical phenomena involved in the The nut is in stainless steel or aluminium.
complete process of installation and loading.
Anchors are installed by driving the anchor sleeve against the
Generally, fasteners can be subdivided into three different locking ring, thus forcing it to expand within the undercut hole
working principles according to the load transfer mechanism, form and locking it within the stone, which provides a stress-
namely friction, bonding and keying. free anchorage under zero applied load. This system is
generally identified with the Fischer-type technology
For friction-type anchors, the tensile load is transferred from (Deutsches Institut für Bautechnik, 2009).
the anchor to the base material due to the friction created by
expanded segments. For bonding anchors, the tensile load is The other system consists of a special anchor made of a
transferred mainly through the adhesive bond between the crosswise slotted anchor sleeve with an internal thread. The
anchor rod and the stone which may cause a combined shear anchor’s upper edge has a hexagon formed to it and the
and cone type failure. respective hexagon bolt with a tooth lock washer formed to it.
The anchor sleeve and the hexagon bolt with a tooth lock
Keying-type anchors carry the tensile load through main keys washer formed to it are also made of stainless steel. The anchor
at the end of the anchor, resulting in a cone shape failure or in is fitted into an undercut drill hole and, by driving the sleeve in
the yielding of the steel rod. This is the case of undercut it is deformed (Figure 2). The anchor sleeve is expanded to its
anchoring. This technology, combined with a suitable support original dimension by inserting the screw to a controlled depth,
framework, allows the engineer to implement a safe and high so that the sleeve sits snugly against the undercut section of the
performance fixing system for stone cladding façades. The hole in the façade panel. This system is identified in general
main aspects of this technology are described in the next with the Keil-type technology (Deutsches Institut für
section. Bautechnik, 2004).

2. Undercut anchoring In each case the anchor embedment depth may vary and is
directly related to the anchor’s tensile capacity. The greater the
Since anchoring is only a small part of the whole façade
embedment depth the higher the load bearing capacity,
system, all other influencing factors must be given an equal
providing the steel has the required load bearing capacity.
amount of consideration for a successful design. However, to
comprehend how undercut anchors are used, a basic under-
standing of the most important issues of this type of The undercut drilling is performed with a proper drill bit and a
technology must be addressed. special drilling device in order to obtain the correct shape and
dimension according to the size and type of anchors to be fitted
(Figure 3).
There are mainly two types of undercut technologies to provide
a keying-type anchorage in the interior of the dimension stone
or slab’s thickness. a

One system is illustrated in Figure 1 in which the undercut

anchors have a cone bolt, either with external thread or b
internal thread and generally 6 or 8 mm in diameter, an
expansion ring with three or four convolutions, a sleeve and,
optionally, a nut. Cone bolts and expansion rings are made of c

a b c a b
d
c
e

Anchor with external thread Anchor with internal thread Figure 2. System using a crosswise slotted sleeve and an internal
thread with a hexagon bolt: (a) slotted sleeve; (b) hexagon bolt
Figure 1. Undercut anchors with external and internal thread: with internal thread and tooth lock; (c) dimension stone; and
(a) expansion ring; (b) sleeve; (c) cone bolt; and (d) nut (d) panel bracket

159

Proceedings of the Institution of Civil Engineers - Construction Materials 2013.166:158-174.

 Construction Materials Undercut anchorage in
Volume 166 Issue CM3 dimension stone cladding
Camposinhos

Stone types and specimens from the same quarry and batch
t
hv were also used to evaluate the force of dowel anchorages and,
du thereby, compare the resistance of these two anchorage
t systems.
Z

u
A semi-empirical formulation is proposed based on results
from testing mechanical stone properties and the geometry of
the anchorage.

Figure 3. Typical cross-section of a stone slab showing undercut

Finite-element computer analyses of the stress states induced in
drilling: wt, diameter of the cylindrical drill hole; wu, diameter of the
the stone panels by the undercut anchorage were also carried
undercut; du, anchorage depth; hv, constant diameter hole depth;
out. These analyses were the basis for calibrating a simple
z, variable diameter hole depth; t, panel thickness
formula that can be used to estimate the breaking load of the
stone at the undercut anchorage.
In general, the undrilled depth (t2du) is equal to or greater
than 0?4t. This may be justified by the fact that the undercut 3.1 Stone type and specimens
anchorage resistance is governed by the stone thickness in A photo with samples of the six studied rock types is shown in
tension under negative wind pressure on façades. For positive Figure 4.
pressures, the majority of resultant forces are transmitted
directly to the frame supporting system and/or the façade A brief lithological description of these stones is a prerequisite
backup structure. to fully understand the study and its results.

2.1 Frame supporting systems 3.1.1 The granites

In both systems, the screw or nut is screwed in until exerting The study included three different granites: Cinzento de
slight pressure on a panel bracket forming a rigid unit with a Alpalhão (C.A.), Pedras Salgadas (P.S.) and Amarelo Real
backup supporting façade system. (A.R.).

The support system for a dimension stone panel in most cases (a) The Cinzento Alpalhão (C.A.) granite is grey with a thin
consists of a suitable framework in aluminium or stainless grain and a very uniform background. Its colour varies
steel. Generally, the framework consists of four brackets somewhat, from light to dark grey. It is a hard natural
attached to the back of the façade panel by means of an stone containing the following essential minerals:
appropriate undercut anchor. These brackets are then placed
into or onto a corresponding continuous horizontal rail.

In most cases the horizontal rails are attached to vertical

mullions which are fixed to the building’s main structure,
structural concrete or masonry.
C.A. P.S. A.R
As for the other fixing systems, it must be possible to adjust
undercut anchorage systems both horizontally and vertically.
Relative movement between the panel and the framework must
also be taken into account.

3. Case study
The main goal of this study is to investigate the relationship
between flexural strength and breaking load at the undercut ML. S.R. ET.
anchorage and also to gain a better understanding of the
undercut anchorage’s rupture behaviour. Hence, several tests Figure 4. Prismatic samples after flexural strength test: C.A.,
were performed to determine the physical and mechanical Cinzento Alpalhão granite; P.S., Pedras Salgadas granite; A.R.,
properties of the different stone types. Pull-out tests were also Amarelo Real granite; M.L., Moleanos Limestone; S.R., Semi Rijo
performed to study the strength behaviour of this anchorage limestone; E.T., Estremoz marble
system.

160

Proceedings of the Institution of Civil Engineers - Construction Materials 2013.166:158-174.

 Construction Materials Undercut anchorage in
Volume 166 Issue CM3 dimension stone cladding
Camposinhos

plagioclase (35%); quartz (30%), microcline (20%) and (UCS). Sixty cylindrical samples were tested in dried condi-
biotite (10%). tions. The test results are shown in Table 2.
(b) The Pedras Salgadas (P.A.) granite is of a light grey
colour with a thin to medium grain and a uniform Flexural strength was determined with three-point load tests.
background. It is predominantly light grey but also has These tests were carried out on 54 prism samples with a
some brownish and white coloured areas. It is also a hard 50 mm 6 30 mm cross-section and on 54 slab samples with a
natural stone containing the following essential minerals: 200 mm 6 30 mm cross-section in accordance with European
microcline (36%); oligoclase (32%); quartz (25%) and Standard EN 12372 (CEN, 2006). Hence, the prism specimens
biotite (6%). had a total length of 200 mm and were placed on rollers with a
(c) The Amarelo Real (A.R.) granite is of a white-yellow 150 mm gap; the slabs had a total length of 400 mm and were
colour, with a slight brownish tonality, and has a medium placed on rollers with a 300 mm gap. The samples were tested
to gross grain and a uniform background. This granite in dried conditions with a gangsawn finish on the lower
varies according to its colour uniformity and grain surface. The bending strength per stone type and the number of
thickness. It is a medium or medium to coarse-grained tests are shown in Table 3.
whitish-yellow to brownish-yellow granite, showing some
porphyroid tendency and a somewhat pronounced The pull-out breaking load in undercut tests mobilises mainly
weathering and incipient foliation. This granite is also the stone’s tensile strength. The tensile strength is a very
known as Amarelo Vila Real. Its essential minerals are: important characteristic which governs the cracking and thus
microcline (32%); quartz (27%); plagioclase (26%); and the rupture.
muscovite (11%).
Tensile strength is formally defined as the tensile stress
3.1.2 The limestones required in order to cause a failure of an unconfined cylindrical
The Moleanos Macio (M.L.) and the Semi Rijo (S.R.) were the or cubical stone specimen, divided by the cross-sectional area
two studied limestones of the specimen perpendicular to the axis of loading. This is the
direct tensile strength; because of the difficulties related to
(a) The Moleanos Macio limestone is whitish-grey to light gripping the specimens, this is a very unusual test.
beige, of an oolitic tendency, calciclastic and bioclastic,
with some dispersed darkish spots. Otherwise, the tensile strength can be found indirectly, that is
(b) The Semi Rijo limestone is white coloured, with thin grain by relying on another type of test. One of these indirect tensile
and a very uniform background. It has some small darker strength methods is the so-called Brazilian test where a circular
spots throughout its surface and may have some slight solid disc is compressed until failure across a diameter; tensile
signs of fossils. It is a soft natural stone and its main stresses perpendicular to that diameter plane are developed; as
variations depend on the amount of darker spots, grain such, compressive loading machines are used in this test. In the
and signs of fossils. Brazilian test a stone’s indirect tensile strength is generally
defined as:
3.1.3 The Marble
The Estremoz Branco Extra (E.T.) marble is white coloured, F
with thin to medium grain and a very discreet bluish coloured
1. sRt ~{ : :
prt
vein. It is a fairly hard natural stone, and its main variations
depend on the vein intensity as well as the type of white where F is the applied force at failure; r is the radius of the
background and structure. specimen (disc); and t is the thickness of the disc.

3.2 Physical and mechanical properties Yet, it must be said that the above formula for determining the
Samples were tested both wet and dry. The wet samples were indirect tensile strength of stone, which has been extensively
soaked in water at 22 ˚C for a minimum of 48 h and a applied in rock engineering and research fields for more than
maximum of 120 h. The dry samples were dried prior to testing 30 years, is erroneous when the disc has a significant thickness
in a humidity-controlled oven at 65 ˚C and gradually cooled to (Yu et al., 2006).
22 ˚C. The samples’ relevant physical characteristics, deter-
mined according to EN 13755 (CEN, 2008) and EN 1936 Another simple procedure to obtain the indirect tensile
(CEN, 2007b), are shown in Table 1. strength is via flexural strength tests (Weibull, 1939). The
flexural strength of a stone specimen is the maximum tensile
The stones were tested in accordance with EN 1926 (CEN, stress when it is about to break. This stress is calculated based
2007a) to determine their unconfined compressive strength on the formula for linearly elastic bodies. Yet, what matters is

161

Proceedings of the Institution of Civil Engineers - Construction Materials 2013.166:158-174.

 Construction Materials Undercut anchorage in
Volume 166 Issue CM3 dimension stone cladding
Camposinhos

Water absorption: % Volumic mass: kg/m3 Open porosity: %

Rock type Stone identification (Figure 4) (EN 13755) (EN 1936) (EN 1936)

Granite C.A. 0?3 2640 0?7

P.S. 0?4 2610 0?8
A.R. 0?7 2610 1?8
Limestone M.L. 2?4 2540 6?3
S.R. 8?0 2190 19?1
Marble E.T. 0?1 2710 0?3

Table 1. Water absorption, volumic mass and open porosity of the

studied stones

the tensile stress value instead of the flexural strength. It is For simple flexure
known that the calculated maximum flexural stress is greater
than the actual stress in the test specimen because, during the
1 h ið{1=mÞ
testing of a stone prism or slab under flexure, a number of
3. sRt ~ ðsRf1 zsRf2 Þ:2ð{1=mÞ : ðmz1Þ2
factors operate to change the stress distribution in the specimen 2
so as to reduce the maximum stress. It follows that the tensile
strength of a stone in the conditions of a flexure test is higher where sRf1 and sRf2 are the flexural strengths of specimens
than in direct testing under tension (Burshtein, 1967). with volumes v1 and v2, respectively.

The fact that flexural strength is physically greater than axial The value of m is given by the following formula:
tension (according to Davidenkov (1947), and Frenkel and
v2 sRf1
Kontorova (1943)) since, during flexure, the maximum tensile 4. m~ln : ln
v1 sRf2
stress is experienced only by a filament on the convex surface
of the specimen whereas during axial tension all points of the
cross-section experience the maximum tensile stress.
Considering the values from Table 3, the tested stone’s indirect
tensile strength can be estimated by means of Equations 3 and
Based on the statistical theory, Weibull (1939) derived relations
4. The results are shown in Table 4.
between the strength of a brittle material under flexure, sRf,
and the axial tension, sRt, if data from specimens of different
3.3 Undercut anchorage strength tests
volumes are known.
Axial tension tests were performed on the slab specimens with a
nominal thickness, t, of 30 mm to evaluate the anchoring strength
Thus, for pure flexure the following relation can be applied:
with Fischer type anchorages. Two external thread M6 or M8
cone bolts with carbon sleeves (see Figures 1 and 6) were used with
1
2. sRt ~ ðsRf1 zsRf2 Þ:ð2mz2Þð{1=mÞ a corresponding nominal diameter drill hole, wt, of 11 and 13 mm,
2 respectively. The corresponding undercut diameters, wu, are of
13?5 and 15?5 mm, with a tolerance of 0?3 mm for both undercuts.
Stone
identification UCS: MPa
All anchorage strength tests were performed with a digital pull-
Rock type (Figure 4) (average) Quantity
off strength tester with a 16 kN capacity (Figure 5). The pull-
Granite C.A. 253 10 off tester is fitted with a load cell and a high resolution large
P.S. 237 10 digital display unit, thus being suitable for measurements with
A.R. 83 10 a resolution of 10 N. The direct tensile force is applied by
Limestone M.L. 92 10 rotating a hand wheel, through a seat ball which ensures axial
S.R. 55 10 and central load application.
Marble E.T. 97 10
In order to limit tested slabs from bending along the unit’s
Table 2. Average unconfined compressive strength of stones three-foot span, they were positioned over a stiff steel plate
studied as per EN 1926 with a 120 mm diameter circular hole with its centre aligned
with the vertical axis of the load cell. The 250 mm 6 250 mm

162

Proceedings of the Institution of Civil Engineers - Construction Materials 2013.166:158-174.

 Construction Materials Undercut anchorage in
Volume 166 Issue CM3 dimension stone cladding
Camposinhos

Flexural strength (F.S.), coefficient of variation (c.v.) and number of tests (qty)
Rock type Stone
identification Prisms flexural Slabs flexural
(Figure 4) strength: MPa c.v.: % Qty strength: MPa c.v.: % Qty

Granite C.A. 21?03 4?25 9 15?72 6?14 9

P.S. 16?82 9?74 9 12?85 8?12 9
A.R. 8?81 8?28 9 6?66 13?31 9
Limestone M.L. 15?90 8?78 9 11?73 7?37 9
S.R. 7?40 7?00 9 4?97 9?10 9
Marble E.T. 18?30 9?94 9 11?32 8?46 9

Table 3. Flexural strength of tested stones according to EN 12372

square steel plate was 12 mm thick and the 120 mm diameter lead to a relevant discrepancy in the anchorages’ bearing
centre hole was sufficiently large not to interfere with the pull- capacities.
off resistance of the undercut anchorages.
4. Results
This procedure limited splitting of the specimens by bending,
Fifty pull-out tests with M6 cone bolts and 81 pull-out tests with
even though splitting occurred in a few cases and the
M8 cone bolts were performed to determine the pullout load failure
consequent results were obviously rejected (Figure 6).
on the six stone types. The typical failure mode was brittle with a
detaching radial cone shape and eventually, spall (Figure 7).
The load was applied at a constant rate of approximately
0?02 kN/s and all specimens were tested after being kept in dry
The diameter of the cylindrical drill hole, wt, the anchorage
storage for over 1 week, without any additional wetting or
depth, du and the slab thickness were measured before the load
drying procedures.
was applied.
Cone bolt anchors of sizes M6 and M8 with external thread
were used. The cone bolt and expansion ring were made of After a cone type failure, the medium length of the spalls and
stainless steel and the sleeve was made of carbon. The anchors their thickness was measured in order to determine the average
were placed in the undercut drill holes whose typical angle of the cone failure surface a as illustrated in Figure 8.
dimensions are shown in Table 5 according to Figure 3. It is
worth pointing out that the difference in the geometry of the The results for the M6 and M8 thread cone bolts are presented
drills for M6 and M8 cone bolt threads is very small – a 2 mm in Tables 6 and 7, respectively. Both tables also indicate the
variation for the diameter of the cylindrical drill hole, wt and coefficients of variation (c.v.) of the breaking load and the spall
for the diameter of the undercut, wu – and probably will not angles.

Flexural strength (F.S.) and estimated tensile strength (T.S.) (mean values)
Stone
identification Prisms flexural Slabs flexural strength: Indirect tensile strength: MPa
Rock type (Figure 4) strength: MPa MPa m (Equation 3) (Equation 1)

Granite C.A. 21?033 15?720 6?946 9?16

P.S. 16?824 12?850 8?495 8?05
A.R. 8?810 6?662 7?037 3?88
Limestone M.L. 15?900 11?730 6?946 6?88
S.R. 7?396 4?967 5?290 2?71
Marble E.T. 18?300 11?320 4?311 5?81

Table 4. Flexural and indirect tensile strength of tested stone

according to Equations 2 and 4

163

Proceedings of the Institution of Civil Engineers - Construction Materials 2013.166:158-174.

 Construction Materials Undercut anchorage in
Volume 166 Issue CM3 dimension stone cladding
Camposinhos

Figure 6. Splitting failure in a limestone slab and undercut

anchorage cone bolt after split failure

Substituting the value of the average spall angle, am, in

Equation 5, the minimum edge distance, em, for anchorage
depths of 15 mm will be given by:

. em ~15|cotð0:839|180 Þ&55 mm
Figure 5. Digital pull-off strength tester and stiff steel plate to
deter bending of the specimens which is in good agreement with the value recommend in the
ETA document (Deutsches Institut für Bautechnik, 2009).

For the M8 cone bolts, since it was not possible to obtain With this minimum space between anchors, sm, may be
samples of Estremoz marble from the same quarry and batch, derived:
there is no available data.
6. sm ~2hv :cotð0:839am Þ
A preliminary observation to the results reveals that the
breaking load does not vary with the thread size for the same
stone type, as was expected. A suitable substitution in Equation 6 makes sm equal to 7?4hv,
which is practically the same as the recommended value in the
For each stone type, a global appreciation of the breaking load Fischer ETA document, 8du (Deutsches Institut für (Bautech-
values and the respective mean value is shown in the graph of nik, 2009).
Figure 9. The gross horizontal lines indicate the average of the
breaking load values shown in Tables 6 and 7. 4.1 Breaking load in the dowel hole comparison
For the same stone types and batch, 10 samples of each type
It must be noted that the spall angles are very similar in all the were used in tests designed to determine the breaking load at
tested specimens, even for the six stone types. The observed the dowel hole in terms of EN 13364 (CEN, 2002). The
mean value for all stone types is of 18?1 ˚ with a coefficient of samples, measuring 200 mm 6 200 mm, were prepared from
variation of 9?84%. These values are slightly inferior to the 20 ˚ slabs with a nominal thickness of 30 mm. The holes were
value reported by Lammert and Hoigard (2007). located 100 mm from either side, measured to the nearest
0?5 mm. The thickness of stone between the edge of the hole
The characteristic spall angle value (5% lower percentile), and the two sides was 12 mm, measured to the nearest 0?5 mm.
assuming a normal distribution, will be given by: The holes were 8 mm in diameter and 35 mm deep. The load
was exerted in a direction perpendicular to the axis of the
5. ak ~am :ð1{1:64|9:84%Þ&0:839am dowel at a maximum distance of 2 mm from the edge of the
sample and using the system shown in Figure 10.

164

Proceedings of the Institution of Civil Engineers - Construction Materials 2013.166:158-174.

 Construction Materials Undercut anchorage in
Volume 166 Issue CM3 dimension stone cladding
Camposinhos

External thread gauge wt: mm wu: mm du: mm z: mm

M6 11 13?5 15 4
M8 13 15?5 15 4

Table 5. Undercut drill geometry for M6 and M8 diameter thread

applied to tested anchorages

The dowel hole failure mode is characterised by a brittle found that the average anchorage capacity of undercut anchors
detachment of a wedge. A sample, after being tested, is was three times greater than that of the dowel hole.
illustrated in Figure 11.
4.2 Semi-empirical formulation
The mean breaking load values at the dowel load are compared In this section, the undercut geometry and the pull-out test
with the mean breaking load values with undercut anchorages results were used to develop a formulation of the maximum
for the six stone types. The results are shown in Figure 12. tensile stress installed in the stone’s anchorage zone.

The comparison is quite clear and, as expected, the anchorage Hence, the spalls created during the undercut cone bolt anchor
capacity with undercut holes is much superior to the dowel tests were traced, digitised and the projected spall area
hole at the slab’s edges. This fact has been already referred to calculated.
by other authors (Camposinhos, 2009; Camposinhos and
Camposinhos, 2009; Lammert and Hoigard, 2007; Stein, By projecting the failure surface in a plane perpendicular to the
2000). In the present case, the difference in percentage varied cone axis and assuming a uniform distribution of the tensile
from 245% for C.A. granite to 78% for S.R. limestone. This strength over an equivalent circular idealised area according to
difference increased with the stone’s tensile strength as revealed Figure 13, it is possible to establish a relation based on the
when compared with the breaking load values with the tensile maximum principal stress theory (Rankine, Lamé) which is
strength values in Table 4. For the six studied types, it was satisfactorily applicable to brittle or quasi-brittle materials,
such as stone. The theory is based on limiting normal stress.
Failure occurs when the normal stress reaches a specified upper
limit. Failure is predicted when the principal stresses equal the
ultimate strength of the material.

The presence of cuts, grooves, holes, etc., modify stress

distributions, which are obtained through elementary stress
design formulas and are based on the members having a
constant section or a section with a gradual contour change, so
that localised high stresses occur and are measured by a stress
concentration factor (Albrecht and Yamada, 1977; Pilkey and
Pilkey, 2007). In this case, a stress concentration factor must be
considered for the given geometry of the undercut drill hole.
This factor, Ku, is defined by the ratio between the actual
maximum stress, smax, and the medium stress, stm according to
Figure 14, Ku is given by Equation 7 as follows:

du
t

Figure 7. Typical cone mode failure in two tested specimens Figure 8. Cone mode failure and spall angle a

165

Proceedings of the Institution of Civil Engineers - Construction Materials 2013.166:158-174.

 Construction Materials Undercut anchorage in
Volume 166 Issue CM3 dimension stone cladding
Camposinhos

Breaking load Spall angle

Stone Number
identification of tests F: kN c.v.: % a: ˚ c.v.: % t: mm wt: mm du: mm

C.A. 5 10?26 2?80 17?79 12?4 31?8 10?9 15?2

P.S. 5 7?71 4?83 19?94 15?0 30?9 10?9 15?2
A.R. 7 3?86 8?92 15?44 9?6 32?9 10?9 15?3
M.L. 8 5?64 22?24 18?62 19?7 29?7 10?9 15?3
S.R. 5 2?34 3?70 19?12 14?0 28?5 10?9 15?3
E.T. 20 4?92 8?02 15?14 13?4 30?7 10?8 15?3

Table 6. Number of tests for each stone type, breaking load and
mean spall angle values and coefficient of variation with M6 thread
cone bolt anchors

smax st :  2 : 2 
7. Ku ~ 10. Ku ~ p hv cot azhv :wu :cota
stm Fu

In Figure 13, the area of the projected cone failure surface, Acf, The chart in Figure 15 shows, for each stone type, the
is represented and may be given by: concentration factor value obtained from Equation 10 with the
h i average values for Fu, a and hv taken from Table 6 and the
Acf ^p ðhv : cot azðwu =2ÞÞ2 {ðwu =2Þ2 diameter of the undercut wu from Table 5. The tensile strength
8.   values from Table 4 were substituted in Equation 10, respectively.
~p h2v : cot2 azhv :wu : cot a
Table 8 reveals that the same stone type showed no relevant
differences in the concentration factor values, which leads to
The breaking load Fu value may then be expressed as follows:
the conclusion that the thread size of the cone bolt anchors and
st : the dimensions of the drilled hole do not affect the said values.
9. Fu ~ Acf
Ku
The medium to gross grain granite (A.R.) shows a higher stress
where hv is the maximum spall thickness equal to (du2z), see concentration factor and a smaller spall angle when compared
(Figures 3 and 13); st is the tensile strength of the stone; Acf is with the other granites, namely the C.A. and P.S. granites.
the area of the projected cone failure surface; and Ku is the Similarly, but obviously for other reasons, the Estremoz
stress concentration factor for the undercut. marble, E.T., which has a large degree of recrystallisation
has a smaller spall angle and a higher concentration factor in
By substituting Equation 8 into Equation 9 and expressing the comparison with the noncrystalline stones such as M.L. and
result for Ku, we obtain: S.R. limestone.

Breaking load Spall angle

Stone Number
identification of tests F: kN c.v.: % a: ˚ c.v.: % t: mm wt: mm du: mm

C.A. 12 10?53 7?22 18?76 22?5 31?7 13?0 13?0

P.S. 25 9?14 12?63 19?23 16?8 31?0 13?0 13?0
A.R. 25 4?42 10?17 15?92 10?8 32?9 13?1 13?1
M.L. 9 5?41 13?85 19?22 21?7 29?3 12?9 12?9
S.R. 10 2?45 18?69 19?93 9?7 28?5 13?0 13?0

Table 7. Number of tests for each stone type, breaking load and
medium spall angle values and coefficient of variation with M8
thread cone bolt anchors

166

Proceedings of the Institution of Civil Engineers - Construction Materials 2013.166:158-174.

 Construction Materials Undercut anchorage in
Volume 166 Issue CM3 dimension stone cladding
Camposinhos

12.00

11.00
(C.A. 10.45 kN)

10.00

9.00 (P.S. 8.90 kN)

8.00

7.00

6.00
kN

(M.L. 5.52 kN)

5.00 (E.T. 4.92 kN)

(A.R. 4.30 kN)

4.00

3.00
(S.R. 2.41 kN)

2.00

1.00 C.A. P.S. A.R. M.L. S.R. E.T.

0.00

Figure 9. Mean anchorage breaking load values per stone type

with Fischerß cone bolt anchors

4.3 Finite-element stress calculation method assumed to be linear-elastic, isotropic and homogeneous in
A finite-element stress analysis was carried out to investigate behaviour, even though dimension stone, as a naturally
stress distribution near the undercut and along the observed occurring material, is often heterogeneous and anisotropic.
spall surface. The material properties used in the model were However, the model was used to verify the induced stress state,
namely the maximum principal stress for the observed pull-out
loads.

A three-dimensional computer model was generated using

solid elements to investigate the stress distribution present in

Figure 10. Testing rig and arrangement for breaking load at the
dowel hole Figure 11. Specimen edge after the test according to EN 13364

167

Proceedings of the Institution of Civil Engineers - Construction Materials 2013.166:158-174.

 Construction Materials Undercut anchorage in
Volume 166 Issue CM3 dimension stone cladding
Camposinhos

12.00

10.00

8.00

6.00
kN

4.00

2.00

0.00
C.A. P.S. A.R. M.L. S.R. E.T.
Undercut anchorage 10.45 8.90 4.30 5.52 2.41 4.92
dowell hole 3.03 2.67 1.36 1.76 1.36 1.98

Figure 12. Breaking load values comparison – undercut cone bolt

anchorage and at dowel hole

the typical undercut anchorage configuration load tested by

Fu
the author, as described above.
hv

t
For this analysis, the eight-noded hexahedron element type
was chosen. Although not relevant for this study, different
values of the modulus of elasticity ranging from 40 000 to
60 000 MPa were used for each type of rock studied. A
hv cot u
constant value of 0?3 for the Poisson ratio was considered
adequate for all rock types.

Acf
tm

max

Figure 14. Stress concentration factor as the relation between the

Figure 13. Spall configuration and detachment angle in slab actual maximum stress, smax and the medium stress stm

168

Proceedings of the Institution of Civil Engineers - Construction Materials 2013.166:158-174.

 Construction Materials Undercut anchorage in
Volume 166 Issue CM3 dimension stone cladding
Camposinhos

10.00

9.00

8.00

7.00

6.00

5.00
Ku

4.00

3.00

2.00

1.00

0.00
C.A. P.S. A.R. M.L. S.R. E.T.

6 mm 4.92 4.69 7.16 6.75 5.62 8.96

8 mm 4.82 4.44 6.32 6.96 5.27

Figure 15. Stress concentration factor, Ku, for the six stone types
studied with M6 and M8 cone bolt undercut anchorages

For simplicity, the load was applied directly to the solid element a volumetric structure modelled using isoparametric volu-
nodes at the nodes in contact with the expansion ring of the cone metric finite elements, of the eight-node brick type (Figure 16).
bolt thread assuming full contact along the stone surface.
Rigid supports were defined at all nodes according to the test
In the finite-element model (FEM), the advantages of the setup. At the nodes belonging to the symmetry face, the
symmetry are put to use. The computational model is taken as degrees of freedom were blocked accordingly.

Concentration
Spall angle a: ˚ factor
Stone identification
M6 cone bolts M8 cone bolts Mean Ku

Fine to medium grain size granite C.A. 17?79 18?76 18?28 4?87
Fine to medium grain size granite P.S. 19?94 19?23 19?59 4?56
Medium to gross grain size granite A.R. 15?44 15?92 15?68 6?74
Limestone M.L. 18?62 19?22 18?92 6?85
Limestone S.R. 19?12 19?93 19?53 5?45
Marble E.T. 15?14 – 15?14 8?96

Table 8. Spall angles and stress concentration factors, mean values

169

Proceedings of the Institution of Civil Engineers - Construction Materials 2013.166:158-174.

 Construction Materials Undercut anchorage in
Volume 166 Issue CM3 dimension stone cladding
Camposinhos

11 mm

3 0 mm
4 mm

mm
7. 3

mm
60

Z
Y
X

Figure 16. Finite-element mesh for a quarter-sample of the studied

slab area

Stress contour maps

Fu = 10 454 N

Figure 17. Three-dimensional stress maps for the C.A. granite

breaking load equal to 10?45 kN and FEM calculated maximum
tensile stress equal to 10?20 MPa

170

Proceedings of the Institution of Civil Engineers - Construction Materials 2013.166:158-174.

 Construction Materials Undercut anchorage in
Volume 166 Issue CM3 dimension stone cladding
Camposinhos

Nodal stresses were averaged from the stresses evaluated at the is desirable to compare the values obtained through both
standard 2 6 2 6 2 Gauss integration points. methods, such that the stress value found in this manner may
be compared with the semi-empirical formulation derived from
Figures 16 and 17 show, respectively, a three-dimensional view of Equation 10 as follows:
the geometry and a stress map, Figure 18 shows a two-dimensional
section cut according to the calculated stone’s maximum principal Fu :Ku
11. st ~  2 : 2 
tensile stresses. Stress concentrations are present in the vicinity of p hv cot azhv :wu :cota
the hole and dissipate quickly, suggesting that anchorage failures
are influenced by a non-uniform stress distribution over the failure
surface. The direction of the maximum principal stresses located A comparison between the maximum calculated tensile stress
adjacent to the hole are approximately perpendicular to the failure using the FEM method and the stone’s indirect tensile strength
surface observed in tests, meaning that failure initiates at this obtained via flexural tests is shown in Table 9. The stress
location, as was expected. values for both cases are also plotted in Figure 19.

One run for each breaking load value was performed with the It must be emphasised that the difference in the stress values
corresponding value for the six different stone types as were 12?6% superior for granites and 12?0% inferior for the
reported in Figure 9. limestone and marble stone types. These stress value differ-
ences may be considered acceptable, considering the variations
in the breaking load test results (Tables 6 and 7) and the
5. Discussion stone’s flexural strength or its indirect tensile strength
The universal availability of powerful, effective computational (Table 4). Nevertheless, note the unavoidable shortcomings
capabilities, usually based on the FEM, has altered the need of the finite-element modelling, in particular the geometry
for and the use of stress concentration factors. Nevertheless, it discretisation and material constitutive law.

Fu = 4924 N
Max stress contours
Radial section cut

Figure 18. Stress maps, two-dimensional view of a radial section

cut of the drilled region for the E.T. marble’s breaking load equal to
4?92 kN and FEM calculated maximum tensile stress equal to
5?04 MPa

171

Proceedings of the Institution of Civil Engineers - Construction Materials 2013.166:158-174.

 Construction Materials Undercut anchorage in
Volume 166 Issue CM3 dimension stone cladding
Camposinhos

Average breaking load values Indirect tensile strength

Rock type Stone identification (Figure 9): kN (Table 4): MPa FEM stress values: MPa

Granite C.A. 10?45 9?16 10?20

P.S. 8?90 8?05 9?10
A.R. 4?30 3?88 4?40
Limestone M.L. 5?52 6?88 5?64
S.R. 2?41 2?71 2?58
Marble E.T. 4?92 5?81 5?04

Table 9. Breaking loads, tensile strength and stress values from the
FEM calculation of tested stone

Finite-element method Tensile strength Semi-empirical (Equation 11)

Cinzento Alpalhão (C.A.)

10.000

8.000

Estremoz (E.T.) 6.000 Pedras Salgadas (P.S.)

4.000

2.000

0.000

Semi Rijo (S.R.) Amarelo Real (A.R.)

Moleanos (M.L.)

Figure 19. Calculated stress with the FEM method and Equation
11 compared with tensile strength for the studied stone types
(values in MPa)

172

Proceedings of the Institution of Civil Engineers - Construction Materials 2013.166:158-174.

 Construction Materials Undercut anchorage in
Volume 166 Issue CM3 dimension stone cladding
Camposinhos

Indirect tensile strength Finite-element method Stress calculated values

Rock type Stone identification (Table 4): MPa stress calculated values: MPa Equation 11: MPa

Granite C.A. 9?16 10?20 9?72

P.S. 8?05 9?10 8?80
A.R. 3?88 4?40 4?18
Limestone M.L. 6?88 5?64 7?70
S.R. 2?71 2?58 2?83
Marble E.T. 5?81 5?04 5?97

Table 10. Comparison between maximum stress from the FEM and
Equation 11 with indirect tensile strength

For the geometry of the undercut drill hole, hv, was regarded as Using the FEM with linear-elastic material properties, together
equal to 11?3 mm and an average value of 14?5 mm was with the maximum principal stress failure criteria, has proven to be
applied for wu. an appropriate design procedure for estimating the breaking load.

To make use of Equation 11, the medium spall angles and Observation of the test results allows for some further
stress concentration factors were taken from Table 8. conclusions.

Table 10 shows the comparison between stresses calculated (i) (a) The breaking load does not vary with the thread size for
using the FEM (ii), Equation 11 and (iii) the indirect tensile the same stone type.
strength. (b) Anchorage embedment depth is directly related to the
anchor’s tensile capacity.
As for the semi-empirical approach in Equation 11, the (c) Based on the semi-empirical formulation for the same
calculated values appear to be conservative. On average they tensile strength, it can be shown that the greater the
are 7?8% superior for the granites and 6?4% superior for the embedment depth the higher the load bearing capacity.
limestone and marble stone types, when compared with the (d) Spall angle measurements made it possible to define a
indirect tensile strength. minimum space between anchors and distance to edges.
(e) For the studied specimens, it was found that the medium
A plot of the three groups of stress values from Table 10 and anchorage capacity of the undercut anchors is three times
per stone type is shown in Figure 19. greater than that of the dowel anchorage type.

6. Conclusion Acknowledgements
This paper examined the relationships between material The author wishes to thank Dr Roland Unterweger for his
strength, anchorage strength and induced stress states for the support and the back-up provided by the Fischerwerke GmbH
undercut anchorage. The findings are based on a significant & Co. KG in supplying the hardware and the specimen
number of tests performed on six stone types, three granites, undercutting drills.
two limestones and one marble.
REFERENCES
Anchorage failure can be described as inducing a cone-shaped Albrecht P and Yamada K (1977) Rapid calculation of stress
spall. The projected area of the stone failure surface is of a intensity factors. Journal of the Structural Division, ASCE
somewhat circular shape. The average surface failure angle was 103(2): 377–389.
found to be of approximately 18 ˚. Burshtein LS (1967) Rock strengths under axial tension and
flexure. Fiziko-Tekhnicheskie Mining Institute Leningrad
The maximum principal stress theory is applied satisfactorily to a 43(4): 53–61.
limiting normal stress. Failure occurs when the normal stress Camposinhos RS (2009) Revestimentos em Pedra natural com
reaches a specified upper limit. Based on this assumption, and on Fixação Mecânica – Dimensionamento e Projecto. Edições
the described failure area and angle, the breaking load value at the Sı́labo, Lisboa, Portugal (in Portuguese).
undercut hole can be estimated with sufficient accuracy based on Camposinhos RS and Camposinhos RPA (2009) Dimension-stone
the indirect tensile strength estimated from flexural strength tests. cladding design with dowel anchorage. Proceedings of the

173

Proceedings of the Institution of Civil Engineers - Construction Materials 2013.166:158-174.

 Construction Materials Undercut anchorage in
Volume 166 Issue CM3 dimension stone cladding
Camposinhos

Institution of Civil Engineers – Construction Materials Deutsches Institut für Bautechnik (2009) European Technical
162(3): 95–100. Approval ETA-05/0266 – Fischer-Zykon-panel anchor
CEN (European Committee for Standardisation) (2002) EN FZP(-W). Special Anchor for the rear fixing of façade
13364. Determination of the breaking load at the dowel panels made of selected natural stones according to
hole, Natural stone, Test methods, CEN – TC 246. CEN, EN 1469. Fischerwerke GmbH & Co. KG, Berlin,
Brussels, Belgium. Germany.
CEN (2006) EN 12372 Determination of flexural strength under Frenkel YL and Kontorova MI (1943) Statistical theory of
concentrated load. CEN, Brussels, Belgium. strength. Fiziko-Tekhnicheskie Problemy Razrabotki
CEN (2007a) EN 1926 Natural stone test methods. Poleznykh Iskopaemykh (Journal of Mining Science). 7:
Determination of uniaxial compressive strength, Natural 108–114 (in Russian).
stone, Test methods. CEN, Brussels, Belgium. Lammert BT and Hoigard KR (2007) Material strength
CEN (2007b) EN 1936 Natural stone test methods. considerations in dimension stone anchorage design.
Determination of real density and apparent density, and of Journal of ASTM International 4(6): 40–57.
total and open porosity, Natural stone, Test methods. Pilkey W and Pilkey D (2007) Petersin’s Stress Concentration
CEN, Brussels, Belgium. Factors. Wiley, New York, NY, USA.
CEN (2008) EN 13755 Natural stone test methods. Stein A (2000) Fassaden aus Natur- und Betonwerkstein:
Determination of water absorption at atmospheric pressure, Konstruktion und Bemessung nach DIN 18516, Verlag
Natural stone, Test methods. CEN, Brussels, Belgium. Georg D.W. Callwey, Munich, Germany (in German).
Davidenkov NN (1947) Statistical theory of the scale effect. Weibull W (1939) A Statistical Theory of the Strength of
Journal of Applied Mechanics 14: 63–67 (in Russian). Materials. Royal Swedish Institute of Engineering
Deutsches Institut für Bautechnik (2004) European Technical Research, Stockholm, Sweden.
Approval ETA-03/0055 – Keil Hinterschnittanker KH. Special Yu Y, Yin J and Zhong Z (2006) Shape effects in the Brazilian
anchor for the fixing of façade plates from their back side made tensile strength test and a 3D FEM correction.
of dust-pressed ceramic plates (stoneware) according to EN International Journal of Rock Mechanics & Mining Sciences
176. KEIL Werkzeugfabrik, Berlin, Germany. 43(4): 623–627.

WHAT DO YOU THINK?

To discuss this paper, please email up to 500 words to the
editor at journals@ice.org.uk. Your contribution will be
forwarded to the author(s) for a reply and, if considered
appropriate by the editorial panel, will be published as
discussion in a future issue of the journal.
Proceedings journals rely entirely on contributions sent in
by civil engineering professionals, academics and stu-
dents. Papers should be 2000–5000 words long (briefing
papers should be 1000–2000 words long), with adequate
illustrations and references. You can submit your paper
online via www.icevirtuallibrary.com/content/journals,
where you will also find detailed author guidelines.

174

Proceedings of the Institution of Civil Engineers - Construction Materials 2013.166:158-174.