Construction Materials Proceedings of the Institution of Civil Engineers

Construction Materials 165 June 2012 Issue CM3

Volume 165 Issue CM3 Pages 145–159 http://dx.doi.org/10.1680/coma.10.00018
Dimension stone design – partial safety Paper 1000018
Received 23/03/2010 Accepted 20/12/2011
factors: a reliability based approach Keywords: design methods & aids/risk & probability
Camposinhos analysis

ice | proceedings ICE Publishing: All rights reserved

Dimension stone design – Rui de Sousa

partial safety factors: a Camposinhos
reliability based approach
1
Digitally signed by Rui de Sousa Camposinhos
DN: c=PT, st=Porto, l=Porto, o=Instituto Superior
de Engenharia do Porto, ou=501540709,
ou=rdc@isep.ipp.pt, cn=Rui de Sousa
Camposinhos, email=rdc@isep.ipp.pt
Rui S. Camposinhos PhD Date: 2012.07.11 18:31:34 +01'00'
Coordinator Professor at ISEP, Polytechnic of Porto, School of Engineering,
Porto, Portugal

Today’s thin stone veneers must be designed to resist, besides their self-weight, high wind pressures and induced
seismic forces. They must accommodate hygrometric differential movement, deflection, vibration and creep, etc.
When available, code requirements do not provide an accurate or realistic safety factor for the specific type of anchor
and or stone used in a project. Global safety factors that are recommended by stone industry associations are
imprecise and used as rules of thumb. Limit state design has replaced the older concept of allowable stress design in
most forms of civil engineering. As a result, all modern buildings are designed in accordance with a code which is
based on limit state theory for all man-made materials, yet there is an astonishing nonexistence of 9natural stone9. In
this paper partial factors of safety are proposed depending on the types and on the coefficients of variation of the
distributions of resistances. Their values are determined using structural reliability analysis for the load and resistance
factor design format according to Eurocode 1990. Aging and stone resistance decay is outside the scope of the article,
yet some guidance is provided on the influence of stone durability on stone cladding performance. An application
example is used to illustrate both methods, and conclusions are drawn.

Notation g ageing factor

FS global safety factor mR mean resistance
kR constant mS mean load effect
kS constant sadm maximum (allowable) stress acting on the
l span length slab
MRd design resisting bending moment per unit of sRk characteristic value of the stone strength
width sstr flexural bending strength from tests.
Pf probability of failure
R resistance or capacity
Rk characteristic resistance
r system reliability
S resulting stress or demand action 1. Introduction
Sk characteristic load effect Dimension stone has its own unique qualities that distinguish it
t slab thickness from ‘man-made’ materials. It is not manufactured; blocks are
VR coefficient of variation for R removed from the quarry, slabs are cut from these blocks and
VS coefficient of variation for S then sawn into the final stone parts to be mounted either with
w service value of the wind pressure face finishing or not. Each block is different and each slab is
X random variable different. A skilful combination or matching of dimension
Xjk characteristic value of the basic resistance stone blocks, veneer panels, tops, etc., beautifully combines
variable natural stone’s variety with man’s design. In contrast to the
Xjd design values of the basic resistance vari- uniformity of materials produced by machines or assembly
able lines, dimension stone’s naturally varied appearance is a
Z random variable magnificent work of art and the term ‘uniformity of material’,
b reliability index if applied to natural stone, has a relative meaning.
cd design factor of safety
cf partial safety factor Today’s thin stone veneers must be designed to resist, besides
ck characteristic factor the self weight, high wind pressures and induced seismic forces.

145
Construction Materials
Volume 165 Issue CM3
Most of the time, the stone must accommodate hygrometric
differential movement, deflection, vibration and creep of the
support concrete structure, in addition to weathering and
deterioration of the actual cladding stones.
When available, code requirements do not provide an accurate
or realistic safety factor for the specific type of anchor and/or
stone used in a project. In fact, safety factors for stone have
always been and are still used as rules of thumb compared with
actual safety factors for ‘uniform’ materials such as steel or
concrete. Testing the mechanical and physical properties of
man-made materials produced under controlled conditions,
such as steel or concrete, are a common procedure. Natural
stone, however, has a broader range of material properties and
thus limits this testing procedure to only some tests, either in
type or quantity. Furthermore, other ‘nature-made’ materials,
such as timber, are much more suitable to this testing context
than natural stone products.
2. State of the art
The design method based on allowable stresses has been and, it
can be said, still is the most commonly used method for
dimension stone. This method, itself, contains no rules for
defining these allowable stresses.
Traditionally, safety factors have been applied by comparing
the ratio of the ‘design’ strength to the ‘design’ stress (load)
with standard or accepted values for the stone and/or the
situation which is being designed. These safety values are
subjective by taking into account the rupture stresses of the
stone itself. In Europe there are very few recommendations
about safety factors in dimension stone design and they all
cover an allowable stress design approach (AFNOR, 2000;
BSI, 2010).
Global safety factors recommended by standards (ASTM,
2010a, 2010b) range from 3 to 12 and are followed by stone
industry associations (ILI, 1998; Marble Institute of America,
2007). Yet it must be said that safety factors are intended to
account for applied load variations; section size variations;
material strength variations; loss of strength through time; and
workmanship errors. The use of safety factors is given in all
engineering practices; the method of determining the factor
size, however, is a subject of ongoing discussion among
architects and engineers.
When comparing equivalent values in modern codes, although
partial safety factors for limit state design (LSD) are inferior to
the so-called allowable stress design (ASD) factors, this does
not necessarily mean that the latter result in greater safety
(Beall, 1990; Camposinhos, 2009; Camposinhos and Cam-
posinhos, 2009, 2012; Ellingwood and Coombs, 1930; Lewis,
2007).
146
Dimension stone design –
partial safety factors: a
reliability based approach
Camposinhos
One reason that stone safety factors are more conservative is
that stone is a natural material and not a closely controlled
manufactured product. The physical properties of stone, even
for the same quarry, can vary widely. Some stones also lose
strength after repeated heating-cooling and freeze–thaw cycles.
Some gain or lose strength when they are saturated.
The durability of any stone type is closely related to weath-
ering, which is a complex interaction of physical, chemical and
even biological processes. Thus, stone cladding durability
assessment is not an easy task. This issue is outside the scope of
the article, yet other references are provided.
Although attempts have been made to improve design methods
based on allowable stresses to obtain a better definition of load
and resistance, these attempts have not succeeded since, in
practice, all the data and results are dispersed. It must be said
that these global safety factors have been incorrect and
persistently misunderstood; these factors are sometimes derived
from mean values of material strength, disregarding their
dispersion against some probabilistic value of loads or actions.
It must be emphasised that the dispersion of a material’s
properties is rather important even for elements considered
similar. The need to use a statistical approach for studying the
problem of structural safety in dimension stone design is obvious.
Over the past five decades, there has been a gradual move
towards LSD or load/resistance factor design for the so-called
‘man-made’ construction materials. Limit state design requires
the ‘structure’ to meet two principal criteria: the ultimate limit
state and the serviceability limit state. A limit state is a set of
performance criteria (e.g. vibration levels, deflection, strength,
stability, buckling, twisting, collapse) that must be met when
the structure is subject to loads or any type of direct or indirect
action. This approach applies factors to each design assump-
tion component in order to accommodate unknowns.
Limit state design has replaced the older concept of allowable
stress design in most civil engineering aspects. As a result, all
modern buildings are designed in accordance with a code based
on the limit state theory. For example, in the UK steel
structures are designed in accordance with BS 5950, and
reinforced concrete structures with BS 8110, both of which are
codes based on the limit state theory. Australia, Canada,
China, France, Indonesia and New Zealand (and many other
countries) utilise the limit state theory to develop their design
codes. In the purest sense, it is now considered inappropriate to
discuss safety factors when working with LSD, as there are
concerns that this may lead to confusion.
In 1975, the Commission of the European Communities
decided on an action programme in the construction sector
Construction Materials
Volume 165 Issue CM3
to eliminate technical obstacles hampering trade and the
harmonisation of technical specifications. Within this action
programme, the European Commission took the initiative of
establishing a set of harmonised technical rules for the
structural design of construction works: the Eurocodes. The
main goals were and still are as listed here.
(a) To provide a common understanding about structural
design between owners, operators and users, designers,
contractors and manufacturers of construction products.
(b) To provide common design criteria and methods to fulfil
the specified requirements for mechanical resistance,
stability and resistance to fire, including aspects of
durability and economy.
(c) To facilitate the marketing and use of structural
components and kits in European Union (EU) member
states.
(d) To facilitate the marketing and use of materials and
constituent products, whose properties are included in
design calculations, in EU member states.
(e) To be a common basis for research and development, as
the Eurocodes offer the opportunity for pan-European
research for their future editions, leading to substantial
savings in research costs.
(f) To allow the preparation of common design aids and
software.
Simply stated, the action programme aims to promote
European civil engineering firms, contractors, designers and
product manufacturers in their worldwide activities and to
increase their competitiveness.
Nearly all the Eurocodes have already been introduced, thereby
establishing a set of common technical rules for the design of
buildings and civil engineering works which will ultimately
replace the individual rules of the various member states. There
are 10 structural Eurocodes covering the design of structures
made of concrete (EC2), Steel, (EC3), composite steel and
concrete (EC4), timber (EC5), masonry (EC6) and aluminium
(EC9). The other Eurocodes cover the basis of structural design
(BS-EN 1990; BSI, 2006), geotechnical design (EC7) and the
design of structures for earthquake resistance (EC8; CEN, 2004).
Astonishingly, ‘natural stone’ has been practically overlooked
by this programme, bearing in mind that current rules in nearly
all, if not all, countries may be considered archaic or at least
obsolete.
American standards apply different factors of safety, by type of
stone, in an attempt to account for the variability of strength
and other mentioned factors. This is not a limit state approach
but has developed due to its pragmatic approach in addressing
many of the unknowns, not only related with the material
Dimension stone design –
partial safety factors: a
reliability based approach
Camposinhos
itself, but also with the fabrication, erection, transportation,
loading and analysis method.
In any case, selecting and testing stone is the critical element in
thin stone façade projects. Hence, civil engineers and architects
do not have suitable guidance and assistance when designing
stonework structural applications according to the limit state
design principle.
2.1 The factors to be weighed
When assessing civil engineering structures, the capability of a
designed system to respond to project requirements or to meet
user demands must be assessed but without overlooking some
basic but fundamental principles.
Although limit state design does not seek to identify the overall
safety factor applicable to each design case, it is often calculated
by engineers for ease of comparison with the traditional
allowable stress design, particularly in areas where limit state
design is applied to ensure safety according to statistical
concepts and, thus, to establish design rules or even a code.
Discussions on basic requirements for this purpose have been
held at diverse instances (Ditlevesen and Madsen, 1996;
Melchers, 1999; Virlogeux and Walraven, 1999; Walraven,
2004). There are principles that should be regarded in a design
code or in establishing design principles. These same principles
are dealt with in the following sections.
3. Risk and reliability analysis
A defined level of risk has to be accepted subject to economic
and social constraints in order to define or to establish the point
at which a system ceases to operate even with a certain degree of
subjectivity – this decision in itself involves uncertainty and
decision makers must balance the benefits against costs.
In fact a system can fail to perform its intended function for
one or more reasons, such as natural hazards or lower
performance than predicted. Failures may even include such
rare events as the collapse of major structures. Although the
assurance of a system’s safety is primarily a task of engineers,
the accepted levels of risk are subject to economic and social
constraints and in this way social issues play an important role
in the analysis of civil engineering systems, because these
systems are more directly involved with the public than are
other engineering systems.
It must be noted that defining a system’s inability to perform
adequately, thus leading to failure, is not an easy task,
particularly in the case of non-reversible failures.
Determining the stage at which a system ceases to perform
adequately is, to a certain degree, subjective.
147
Construction Materials
Volume 165 Issue CM3
To analyse a system’s risk of failure, one must clearly identify the
input to the system and its consequent response. In the case of a
building, structural safety depends on the maximum load that may
be imposed during the building’s lifetime, and also on the load-
carrying capacity, or strength, of the structure or its components.
This is a general problem which is also related to dimension stone
cladding design with mechanical anchorage. In fact, predicting the
maximum load and actual strength of stone cladding is subject to
the same uncertainties; one cannot ensure its absolute safety, and
engineers have to rely on some probabilistic concept indicating the
likelihood that the available strength will adequately withstand the
maximum load over the lifetime of the façade cladding system.
The said reliability is defined as the probabilistic assessment of
the likelihood that a given system will perform adequately for a
specified period of time under known operating conditions or,
more simply stated, a system’s reliability is defined as the
probability of non-failure during a system’s specified lifetime.
The risk, on the other hand, is defined as the probability of
failure under the same conditions.
As such, the risk of a system’s inability to meet the respective
demand is defined as the probability of failure, Pf, during the
specified system’s lifetime under specified operating conditions.
System reliability, denoted by r in Equation 1, is the
(complementary) probability of non-failure:
1. r~1{Pf
A system’s capability to perform under given requirements can
be defined in the present case using the terms ‘load’ or ‘action’
[It is preferable to use the term ‘action’ instead of ‘load’ since
any load is always an action and the opposite is not totally and
properly applicable.] and ‘strength’ or ‘capacity’.
3.1 Measures of reliability
As already mentioned, assessing risk and safety in cladding
design is traditionally based on ‘allowable factors of safety’;
these are estimated from previous experience regarding the
behaviour of a particular fixing system or from observed
behaviour of similar systems. A common safety factor measure
applied by designers is the ratio between the assumed nominal
values of the material strength, say resistance, R*, and action, S*:
R
2. F ~
S
For example, if the allowable stress in a dimension stone slab is
4?0 MPa and the design stress, due to wind load, is 3?0 MPa, the
conventional safety factor is 1?33. The engineer may assume that
the designed slab thickness is satisfactory if the calculated safety
148
Dimension stone design –
partial safety factors: a
reliability based approach
Camposinhos
factor is greater than an accepted minimum value, which can be
based on experience or on design-imposed prescriptions. If a
safety factor of 1?33 is regarded as low, the engineer should
redesign the slab through two different procedures: increasing the
thickness and thus its capacity; decreasing the distance between
supports and, consequently, reducing induced wind stress.
Because the nominal values of both resistance R* and action S*
cannot be assessed with certainty, the capacity and demand
functions, R*, S* must be considered as probability functions.
Hence, the safety factor, given by the ratio F5R/S of two
random variables ,R ; S. is also a random variable.
The system’s inadequacy to meet the required demand,
measured by the probability of failure, is associated with the
portion of the safety margin distribution whereby it becomes
less than zero. That is, a given structural element will be
considered to have failed if its resistance or capacity, R, is less
than the resulting stress or demand action, S. The structural
element’s failure probability can be stated as follows:
3. Pf ~P ðRƒS Þ~P ððR{SÞƒ0Þ
Defining fS(s) and fR(r) as the probability density independent
of the functions of demand S and capacity, R, the probability,
Pf, of system failure is given by:
ð s§r
? ð
4. Pf ~P ðRƒS Þ~ fR (r)fS (s) dr:ds
{? {?
taking into account that for any random variable, X, the
cumulative distribution function is given by:
ðx
5. Fx (x)~P ðX Þƒx~ fx (y) dy
{?
Provided x>y, it follows that for the common but special case
when S and R are independent, that Equation 4 can be written
in the single integral form, since there is no physical meaning in
defining the distributions to negative values of x:
?
ð
6. Pf ~PðRƒSÞ~ FR (x)fS (x) dx
0
This is also known as a ‘convolution integral’ whose meaning
may be easily explained by reference to Figure 1. FR (x) is the
probability that x>y or the probability that the actual
Construction Materials Dimension stone design –
Volume 165 Issue CM3 partial safety factors: a
reliability based approach
Camposinhos

ðm {mS Þ
resistance R of an element or member is less than a specific less than the mean at a distance higher than qRffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi,
value, x, thus representing failure. By representing all values of s2R zs2S
measured in standard deviations.
x, namely by taking an integral over all x, the total failure
probability is obtained. This is also shown in Figure 2 where
the density functions fR and fS are drawn.
3.2 Reliability index
Consider Equation 9 rewritten in the following form:
For only some distributions of R and S it is possible to integrate the  
convolution integral in Equation 6 analytically. An example is mZ
10. Pf ~W { ~Wð{bÞ
when both are normal random variables with means mRand mS and sZ
correspondingly variances s2R and s2S . The safety margin given by
Z5R-S has a mean and variance given by well-known rules for where b is the so-called reliability index, the most frequently
addition (subtraction) of normal random variables: used reliability measures of structures, for example, BS EC
1990 (BSI, 2006). It should be emphasised that this reliability
7. mZ ~mR {mS index is completely equivalent to the probability of failure Pf
whatever method is used for its calculation, for which it may be
explicitly expressed:
8. s2Z ~s2R zs2S
11. b~{W{1 ðPf Þ

Substituting Equations 7 and 8 in Equation 3 one has:

where b is defined as the inverse distribution function of the
 
0{mZ standardised normal distribution of the probability of failure.
9. Pf ~P(R{Sƒ0)~P(Zƒ0)~W Note that the negative sign on the right-hand side of Equation
sZ
11 is introduced to keep b positive for Pf values of less than 0?5.
W() is the very well-known standard normal distribution from
statistics texts. The random variable Z5R2S is shown in It follows from Equations 9 and 11, based on the assumption
Figure 3 in which the failure region is Z#0.
of a normal distribution for the load effect, S, and resistance,
R, that the reliability index is given by:
It must be remembered that the probability of failure is the
probability of R2S taking negative values. Hence, using mZ
Equations 7, 8 and 9, it follows that this probability is equal to
12. b~
sZ
the probability of obtaining, in a normal distribution, values

FR(x)fs(x)

1.0

FR(x)

fs(x)

x
0.0

P(R x)
R=x

Figure 1. Basic R–S problem: FR(x), fS(x) representation

149
Construction Materials Dimension stone design –
Volume 165 Issue CM3 partial safety factors: a
reliability based approach
Camposinhos

fR(r)fs(s) Load effect - S Resistance - R

e.g. bending moment e.g. flexural capacity

Failure
density

Area = FR(x)fs(x)dx

FR(x)fs(x) x

x x+dx

Figure 2. Basic R–S problem: fR(x), fS(x) representation

The above result is valid only under the limiting assumptions
of the normal distribution of both fundamental variables R
and S. In a more general case, when R and S have a general
‘non-normal’ distribution, the probability of failure Pf cannot
be determined using Equation 11, but only as the first
estimates. In this case, however, the probability of failure Pf
may be still determined following Equation 6 and/or any
specialised software. Although many theoretical procedures
with various degrees of complexity have been developed over
the last few decades, they are not popular with practicing
engineers. One reason for this unpopularity could be the lack
of user-friendly software, since structural engineers without
formal education in reliability-based design may not be able to
use these computer programs.
Let us consider again the dimension stone slab from the
previous example in which the lateral action-induced stress at
mid-span has a mean value of mw53?0 MPa and, for example,
a variance of s2w ~1:0 MPa. The bending strength of similar
slabs has been found to have a median value of mR54?0 MPa
with a coefficient of variation of VR 5 20%. It is preferable to
evaluate the probability of failure, Pf. The slab self-weight is
not relevant for this effect and thus ignored.
As such, the resistance variance is given by:
. s2R ~ðVR :mR Þ2 ~ð0:20|4:0Þ2 ~0:64 MPa

Failure Safety
Z<0 Z>0

z z
fz(z)

Pf
Probabilty of failure

z
0 z

Figure 3. Distribution of safety margin Z 5 R–S

150
Construction Materials
Volume 165 Issue CM3
Hence: mZ5mR2mS542351 MPa and
. s2Z ~s2R zs2S ~0:64z1:0~1:64 MPa:
mZ 1
Therefore b~ ~ &1:2806(3) and, according to
sZ 1:64
Equation 10, the probability of failure is as follows:
. Pf ~Wð{bÞ~Wð{1:28063Þ&10:0%
3.3 Reliability differentiation
The differentiation of the degrees of reliability depends on the
cause and mode of failure, the possible consequences of failure,
the public’s aversion to failure and the expense and level of
effort necessary to reduce the risk of failure. It is obvious that
an element which would be likely to collapse suddenly and
without warning should be designed for a higher degree of
reliability than one whose collapse is preceded by some kind of
warning thus permitting some measures to be taken to limit or
avoid consequences. Similarly, there are issues related to the
consequences of failure such as the risk to life, injury, potential
economic losses, etc., that lead to a different degree of
reliability.
For reliability rating purposes, Annex B of BS EN 1990 (BSI,
2006) stipulates three classes depending on the failure described
in Table 1.
In Eurocode 1990, three reliability classes – RC1, RC2 and
RC3 – are associated with the three consequence classes
described in Table 1. A link is established between these classes
and values for the reliability index, b, concerning ultimate limit
and serviceability limit states for 1 year and 50 year reference
periods. Table 2 illustrates consequences and reliability classes
and values for the reliability index.
As was shown (Equation 10) the reliability index is a function
of the probability of failure. This relation is given in Table 3.
4. Safety factors and characteristic values
The traditional deterministic measures of limit state ‘violation’,
namely the factor of safety and the load factor, can be related
directly to the probability Pf of limit state violation.
Generally, some upper range value of applied load or stress is
compared with some lower range value of material strength.
Such values might be termed ‘characteristic’ values, reflecting
that in conventional usage (e.g., in design) the load or strength
is described only by this value. Thus, for instance, the
characteristic tensile strength of dimension stone slabs is the
strength that most (say 95%) slabs will exceed. There is a finite,
Dimension stone design –
partial safety factors: a
reliability based approach
Camposinhos
yet small, probability that some dimension stone slabs, for
instance, will have an inferior strength.
For resistance, the so-called characteristic values, Rk, are
defined on the low side of the mean resistance:
13. Rk ~mR :ð1{kR :VR Þ
where Rk is the characteristic resistance, mR is the mean
resistance, VR is the coefficient of variation for R and kR is a
constant. This is based on the normal distribution in which Rk
is the value of resistance below which only, say, 5% of slab
samples will fail. Also, for the standardised normal distribution
function it follows that:
 
Rk {mR
14. 0:05~W {
sR
ðmR {Rk Þ
and for a 5% ‘one-sided tail’, k0:05 ~ &1:645.
sR
Expression 14 now follows directly, noting that the standard
deviation can be expressed as sR ~mR :VR .
Similarly, the characteristic value for the load effect is
estimated on the upper side of the mean:
15. Sk ~mS :ð1zkS :VS Þ
where Sk is the characteristic load effect, mS is the mean load
effect, VS is the coefficient of variation for S and kS is a
constant.
If design values are defined, for example, wind velocities, as not
exceeding 98% of the time, a load effect is applied, then
kS<2?0537 if S is normally distributed.
In codified design, the percentiles used (such as 5 and 98%
above) either are explicitly specified or may be deduced from
the characteristic value specified in existing codes or docu-
ments. Other percentile characteristic values can be obtained in
the manner indicated above for normal distributions, and also
for non-normal distributions.
The characteristic factor, ck, is defined as the lower 5% fractile of
the ratio of FR(x) and the upper 5% fractile of FS(x). The design
factor of safety, cd, may be defined as the ratio between the
lower 5% fractile of FR(x) and the upper 5% fractile of FS(x).
Through Equation 6 it is possible to relate the safety factors
with the probabilities of failure. This relation depends on the
types and on the coefficients of variation of the distributions of
resistances and load effects. For determining these relations,
several types of distributions may be considered. This is the
151
Construction Materials Dimension stone design –
Volume 165 Issue CM3 partial safety factors: a
reliability based approach
Camposinhos

Consequence class Description Example for building works

CC1 Low consequence for loss of human life, or
negligible economic, social or environmental
consequences
CC2 Medium consequence for loss of human life, or
considerable economic, social or
environmental consequences
CC3 High consequence for loss of human life, or
very important economic, social or
environmental consequences
Agricultural buildings, sheds, greenhouses, namely,
construction or buildings where people do not
normally enter
Construction for which the consequences of failure are
significant, for example, apartment or office buildings,
hotels, schools, access bridges, etc.
Construction for which the failure consequences are
severe, stadiums, grandstands, theatres, high-rise
buildings, bridges, dams, etc.

Table 1. Consequence class according to EC 1990 and examples

basis in structural reliability analysis of the load and resistance
factor design format (LRFD).
5. Design situations for stone cladding
According to BS EN 1990 (BSI, 2006), after having determined
a design working life for a structure or structural component,
the action variations and environmental influences during that
period, the structural and material properties must be selected
by taking into account distinct situations within a certain time
interval implying inherent hazards or conditions.
5.1 Limit state design
The design procedure using the limit state concept consists of
setting up structural and load models for relevant ultimate and
serviceability limit states which are taken into account in the
various design situations and load cases in order to verify that
no limit is exceeded when relevant design values for loads or
actions and for material or product properties and for
geometrical properties are used in appropriate structural and
load models.
In this context and for the present case, the following design
situations may be regarded as relevant for dimension stone
design.
(a) Persistent situations generally related to the design
working life of the element in normal use, including
extreme loading conditions from wind, snow, imposed
loads, etc.
(b) Exceptional seismic situations applicable to elements
when subjected to seismic events require a design for
protection against earthquake loads.
5.2 Serviceability limit state design
The structural design criteria used for the serviceability limit
state design are normally based on the limits of deflections or
vibration for normal use. In reality, excessive deformation of a
dimension stone slab is normally caused by bowing, and
excessive vibration or noise is caused by a defective anchorage
construction technique. Certain interrelationships may exist
among the design criteria defined and used separately for
convenience purposes. Criteria are normally defined by
established practice and economical in-service performance
without excessive routine maintenance or downtime (ASTM,
2010a, 2010b; BSI, 2010; Gere, 1988; Lewis, 1995).
The acceptable limits necessarily depend on the type, mission and
arrangement of the backup structure. Furthermore, in defining
such limits, it is normally found that they are less important than
the observation of good practices in construction procedures. As

Ultimate limit states Serviceability limit states

Consequences 1 year reference 50 year reference 1 year reference 50 year reference

class Reliability class period period period period

CC1 RC1 4?2 3?3 2?9 1?5

CC2 RC2 4?7 3?8
CC3 RC3 5?2 4?3

Table 2. Differentiation for b values according EC 1990

152
Construction Materials Dimension stone design –
Volume 165 Issue CM3 partial safety factors: a
reliability based approach
Camposinhos

Pf 1021 1022 1023 1024 1025 1026 1027 1028 1029

b 1?282 2?326 3?090 3?719 4?265 4?753 5?199 5?612 5?998

Table 3. Relation between probability of failure, Pf, and reliability

index, b

an example, the limiting values of vertical deflections for slabs in
horizontal planes are far from being critical. In fact, good
judgment by architects, engineers and contractors when specify-
ing, designing, engineering and constructing stone and other
works that interface with stone is indispensable in order to
combine the stone’s known performance characteristics, the
building’s structural behaviour and knowledge of materials and
construction methods with proven engineering practice.
5.3 Ultimate limit state design
Ultimate limit state design or load/resistance factor design for
dimension stone may be applied following a suitable stress
analysis taking into consideration all the relevant factors
involving the design life cycle of stone cladding.
basic variables. Usually the lower value of a material property or
product is unfavourable, and the 5% (lower) fractile is then
considered as the characteristic value. In the case of dimension
stone it can be assumed that the theoretical model for the
random behaviour of its property is known, or sufficient data
may be available to determine such a model. In this case basic
statistical techniques for the determination of the characteristic
value of stone properties may be indicated: For the flexural
resistance a log-normal distribution is considered appropriate to
approximate the data from the tests, and the characteristic value,
Rk, may be obtained as follows:
  
16. Rk ~ exp xln { kp,5% |sln

In the ‘load’ side wind and seismic forces must be assessed  ln , and standard deviation, sln, of
With the logarithmic mean, x
according actual regulation and codes such as EC1 (CEN, the n tested specimen:
2002) and EC8 (CEN, 2004).
1X  
17. x
ln ~ ln xj
For resistance, a set of design rules should be based on clear and n j
scientifically well founded, consistent and coherent theories
corresponding to a good representation of the structural
behaviour and of the respective materials’ physics. It should be sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
P 2
simple enough to be handled by practitioners without considerable ln xj { xln
18. sln ~+
problems and lead to conservative and robust designs. However, n{1
in a number of cases and as an alternative, more detailed design
rules may be offered that consume even more calculation time, but
also result in a more economical design. There are some studies Parameter kp,5% is a coefficient whose value depends on the
applying this strategy for designing dimension stone cladding number of tested specimens.
using dowel, type 31, and kerf anchorages (Camposinhos and
Camposinhos, 2009, 2012; Conroy and Hoigard, 2007; Lammert Use of the normal distribution is generally accepted if the
and Hoigard, 2007; Lewis, 1995; Naggatz and Gerns, 2007; quantity of the tested sample is greater than 50, otherwise the
Scavuzzo and Acri, 2000). These studies have shown that a design CEN committee recommends the adjustment of a log-normal
focused on the bending strength of stone panels, to the detriment distribution (CEN, 2006, 2008).
of the anchorage zone, is an unsafe and yet common practice. All
these studies conclude that designing stone cladding systems must
6.1 Partial safety factors
take into account different effects in order to evaluate the effective
stress in the critical region of the anchorage geometry. In these For ultimate limit states, the fundamental combination of the
studies, separate stress concentration factors were proposed to effect of actions takes the following well-known equation:
account for the anchorage zone geometry and the natural stone’s !
m n
specific properties.
cG :Gk 00 z00 cQ : Qk 00 z00 y0j :Qkj
X X
19. Ed ~
i~1 i~1
6. Dimension stone properties
To determine the structural reliability, the properties of materials where Gk and Qk refer to the characteristic values of the actions
together with geometrical data are an important group of the (permanent and variable) and y0,j is the coefficient defining the

153
Construction Materials
Volume 165 Issue CM3
combination weight value for an action’s particular effect. In
general, for lateral actions affecting cladding, such as wind
pressure and seismic forces, the coefficients of combination y0,j
are zero and the only permanent load is the slab’s self weight.
In this case Equation 19 is expressed as follows:
20. Ed ~cG :Gk 00 z00 cQ :Qk
It must be noted that the plus sign in the above equations can
only be numerical and applied if the effect of the actions are
additive for the design variable under analysis.
The safety coefficients are intended to take into account
uncertainties associated with the analysis model and with the
actions’ intensity variation. They are usually defined in the
structural codes for persistent and transient design situations,
assuming, for the most common cases of building structures
and for unfavourable effects, the consensual values of 1?5 for
cG and the values of 1?5 or 1?35 for cG depending on the
material and the applicable specification in the Eurocode suite.
The partial factors cj for the basic variables Xj, having a
favourable effect on structural reliability (on the resistance
side), may be determined as the relation:
Xj k
21. cj ~
Xj d
where Xjk and Xjd are the characteristic and design values of
the basic resistance variable.
Optimisation methods are available to determine partial safety
and load combination factors corresponding to a predefined
safety level. For simple design situations, such as for a
permanent and a variable load, this is an easy task when
normal distribution suitably represents the basic variables.
As mentioned earlier, reliability-based design features should be
added to commercially available deterministic software which
may include techniques such as a fast convolution and curvature-
based adaptative importance sampling. Among several packages,
CodeCal (Faber et al., 2004) is a programme made available by
the Joint Committee on Structural Safety (JCSS) based on the
structural reliability analysis and on the LDRF format. Through
structural reliability methods, the safety formats of the design
codes – namely design equations, characteristic values, partial
safety factors and load combination factors – may be chosen
such that the reliability level of all structures designed according
to the design codes is homogeneous and independent of the
choice of material and the prevailing loading, operational and
environmental conditions. This process, including the choice of
154
Dimension stone design –
partial safety factors: a
reliability based approach
Camposinhos
the desired level of reliability or ‘target reliability’, is commonly
understood as ‘code calibration’.
Thus, for natural stone cladding partial safety factors may be
calculated for each consequent class described in Eurocode
1990 and assuming the corresponding probability of failure
from EC 1990 in Table 3.
The effect of lateral actions, wind pressure and seismic forces
are assumed to be approximated by Weibull distribution with a
coefficient of variation equal to 40%. (BSI, 2006; CEN, 2002;
Lammert and Hoigard, 2007). For flexural and tensile strength
stresses, a log-normal distribution is considered.
Under these assumptions, the partial coefficient of safety for
natural dimension stone design depends on the coefficient of
variation of tensile strength and dimension tolerances for internal
and external cladding according to EN 13373 (CEN, 2003).
The values of the partial safety factor, cM, described in Table 4,
were determined with the CodeCal algorithm according to
(Faber and Sorensen, 2003; Vrouwenvelder, 2002) recommen-
dations under the above–mentioned considerations for differ-
ent coefficients of variation VR of stone properties and the
three consequence classes mentioned in Table 1.
It is accepted that, in the majority of situations, there are low
consequences for loss of human life and low or negligible
economic, social and environmental consequences. However,
Table 4 provides partial safety factors for all three levels, CC1 to
CC3, since the risk needs to be evaluated and implicitly
understood by the designer when selecting a reliability index. A
designer not familiar with the difficulty and unpredictability
underlying the nature of stone design will find it difficult to
decide on an acceptable probability of failure. The coefficient of
variation, if insufficient testing was performed, can be the first
step during which designers make an error in their design
assumptions. The designer must be aware that it’s very unlikely
that stone cladding failure will cause high consequences for loss
of human life or important economic, social or environmental
consequences. Such is very unlikely because this class of
consequences refers to the collapse of special structures and the
coefficients of variation of the natural stone’s properties, such
that the partial safety factors for these applications (consequence
class of CC3) are far from being plausible and affordable.
It must be emphasised that general situations correspond to a
consequence class of CC1 and are given a 50 year reference
period as the corresponding target index. As such, the
following formula is applicable: b(50)53?3 with a probability
of failure of Pf50?5/1000. This value is very close to the
recommendation by Faddy et al. (2003).
Construction Materials Dimension stone design –
Volume 165 Issue CM3 partial safety factors: a
reliability based approach
Camposinhos

Class of consequence
Coefficient of variation of
stone properties VR CC1 CC2 CC3

, 0?1 1?50 1?88 2?40

0?1 to 0?2 2?40 3?48 5?20
0?2 to 0?25 2?90 4?25 6?60
0?25 to 0?30 3?80 5?90 9?80
0?30 to 0?35 4?80 8?20 14?60
0?35 to 0?40 6?10 10?60 19?40

Table 4. Partial safety factors for dimension stone cladding for

three classes of consequences according to EC 1990

7. Partial safety factors where sRk is the characteristic value of the stone strength; cM is
the partial safety factor from Table 4 and g is the ageing factor
7.1 Ageing and stone resistance decay applied to stone resistance decay.
It must be taken into consideration that the physical properties of
stone differ widely between stone groups and even within the
7.3 Example of application
same stone type, and thus it is not evident that all granites are
In the example, a simple case for calculating the minimum
more ‘resistant’ than marbles and that the latter are more resistant
thickness at mid-span of a dimension slab from a fine-sized
than limestone. In fact, the mineral composition, textural
grained granite is presented using both approaches, namely the
differences, varying degrees of hardness and pore/capillary
ASD and the LSD. The slabs span 900 mm kerf anchored in
structure are the main reasons why the same stone shows the
two opposite edges for a service wind load of 1?9 kPa.
same and uniform signs of alteration (Miglio et al., 1999). These
minerals can be broken down, dissolved or converted into new
The flexural strength was determined with three-point load
minerals by a variety of processes such as frost action, thermal
tests carried out on several prism samples. The bending
expansion, wetting and drying, salt decay, etc.
strength and the number of tests are shown in Table 5.

The decay observed in porous stones applied in building

façades is caused, not only by the petrophysical properties of
the stone and climatic conditions, but also by the location of
the materials on the façade itself and the positioning method.
As mentioned before, durability assessment is beyond the scope of
this article. Nevertheless, some references based on empirical
studies are provided. These studies have been carried out to obtain
coefficients that take into account the resistance decay of different
types of stone (Bernabéu and Cura, 2007; Grelk et al., 2007;
Miglio et al., 1999; Prikryl and Smith, 2007). The resistance decay
of each type of stone may be addressed in the limit state design
approach by multiplying the stone resistance obtained from
laboratory tests by an ageing factor, g, less than 1, depending on
the expected lifetime of the façade cladding, environment exposure
and type of stone (Hansen et al., 2003; Lemvig et al., 2002).
7.2 Design value of the stone strength
The design value of the stone strength may be thus expressed
according to Equation 22.
sRk
22. sRd ~g:
cM
7.4 Allowable stress design
The minimum slab thickness can be obtained from the
allowable stress criterion after applying a global safety factor
to the ultimate strength obtained from the three-point bending
test as given by the following equation:
sstr
23. sadm ~
FS
where sadm is the maximum (allowable) stress acting on the
slab; sstr is the flexural bending strength from tests; and FS is
the global safety factor.
As in American standards, a factor of safety may be applied
per type of stone in an attempt to account for the variability of
strength (among other factors). The global safety factor, FS, is
thus determined accordingly (Gere, 1988; Miglio et al., 1997).
For the coefficient of variation and the type of the tested stone,
the value obtained was FS 5 8. The wind-induced stress at the
slab’s mid-span cross-section corresponding to the service
working load is given by:
3|w:l 2
24. sact ~
4|t2

155
Construction Materials Dimension stone design –
Volume 165 Issue CM3 partial safety factors: a
reliability based approach
Camposinhos

where w is the service value of the wind pressure; l is the span 27. sRk ~sstr :ð1{1:64|VR Þ
length; and t is the slab thickness.

Making Equation 24 equal to Equation 23 the minimum Substituting the corresponding values in Equation 27 we have:
required thickness for the slab is given by:
sRk ~sstr :ð1{1:64|VR Þ
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi .
3|w:FS ~9875|ð1{1:64|0:209Þ~6490:245 kPa
25. t~l :
4|sstr

The design value of flexural strength, sRd, for the tested stone
Substituting the corresponding values into Equation 25 gives: is obtained according to Equation 22:

. sRk 6490:245
. sRd ~g: ~1| ~2238:016 kPa
2: 9
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
3|w:FS 3|1:9|8
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi cM
t~l : :
~ 0 9| ~ 0:0306 m
4|sstr 4|9875
The design resisting bending moment per unit of width, MRd,
that is to say, around 31 mm. at the slab’s mid-span cross–section will depend on the slab’s
thickness, as follows:
7.5 Limit state design
Different factors are used for different actions (loads) to
t2
reflect the varying degree of uncertainties associated with 28. MRd ~sRd
6
determining magnitudes and effects. In the case of the wind
dynamic pressure, according to the Eurocode suite, the
partial safety factor, cf, for variable actions must be assumed
The limit state design equation must be applied:
with a value equal to 1?5. Thus, considering the working load
as the characteristic wind pressure, the design bending
29. MRd §MSd
moment per unit of width in the slab’s mid-span cross-
section is given by:
Thus, by substituting Equations 26 and 28 into Equation 29,
26.
the minimum thickness for the slab is obtained by:
w:l 2
MSd ~cf : ~0:1875|1:9|0:92 ~0:288563 kN:m=m
8
0:288563|6
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
. t§ ~0:027814 m;
2238:016
The design flexural strength of the slab is obtained by dividing
the characteristic value of the flexural strength by the that is to say, around 28 mm.
correspondent partial safety factor for the stone according to
Table 4 for the relevant class of consequence. Although it may be discussed whether the 3 mm difference in
the calculated thickness may be overlooked, there is a
Given that the quantity of tested specimens is reasonably fundamental difference in the manner in which each thickness
adequate, the characteristic value of the flexural stress is was determined. The issue arising from the ASD approach is
obtained assuming a normal distribution for the tested stone that the designer is looking for a ‘spot’ in a sea of possibilities
values. Thus the characteristic stress value for flexural strength, that he cannot assess to determine the level of risk associated
sRk, may be obtained as follows: with its ‘solution’. In the other calculation method using the

Flexural strength (sstr), coefficient of variation (c.v.) and number of tests (Qty)

Stone FS: MPa c.v. Qty

Granite 9?875 20?9% 51

Table 5. Flexural strength of tested stone

156
Construction Materials
Volume 165 Issue CM3
limit state design, the solution is not unique and the designer
can decide or choose different solutions (thickness) according
to each situation while knowing the associated level of risk.
8. Conclusion
In Europe there are very few recommendations about safety
factors in dimension stone design and they all cover the ASD
approach. Like American standards, they apply different safety
factors per type of stone in an attempt to account for the
variability of strength since this is a practical method of
addressing many of the unknowns and the method of analysis.
In current practice, stress checks at mid-span are generally the
only calculation carried out to determine the slab’s thickness.
Applying a limit state approach requires defining this
variability and uncertainties by frequency distribution curves,
which can be achieved through simple and affordable
laboratory testing. On the other hand, a more rigorous
analysis must be carried out to compute the induced stresses
in the anchorage support areas. Although this is not a
straightforward task, research work has been carried out to
validate formulae through computational analysis that permits
validating specific formulae for the stress analysis of different
anchorage systems. Application examples of the indicated
design format have been published (Camposinhos and
Camposinhos, 2009, 2012) with the following concluding
remarks.
In fact, global safety factors based on allowable stresses
currently used to simplify the design appears to insufficiently
cover all aspects required for a more accurate analysis. In fact,
only in some cases where bending stresses prevail (slabs
spanning more than 1400 mm), the calculated overall thickness
at mid-span appears to be determinant.
In these studies, the scatter of the tension or flexural test results
and stress concentrations in the support areas are pointed out
as the main reasons for unsuccessful installations that have
shown to be dangerous or faulty.
It is well known that weathering negatively influences stone
resistance. The consideration of this matter on probabilistic
based reliability methods is not an easy task and requires
further studies. Due to considerable variability of the basic
variables (e.g. environmental influence, actions and stone
types), this may be empirically addressed by applying adequate
coefficients to the stone strength, as referred by other authors.
A limit state approach is proposed and many of these
individual factors are being addressed, either explicitly in the
design or accounted for in the load and resistance factors. The
generalisation of a reliability-based approach is implicit in all
countries for all construction materials except for stone.
Dimension stone design –
partial safety factors: a
reliability based approach
Camposinhos
Selecting an appropriate value of b or Pf, that is, the risk
level, is not an easy matter. A priori values for the structural
probability of failure, either during the planned lifespan or per
annum, are not readily available. In this paper, some guidance
measures in the Eurocode suite were followed taking into
account the lower class of consequences.
The principles outlined above have been applied successfully for
determining a partial factor safety-checking format for design
codes applicable to elements of building structures. Some
difficulties were faced in extending the application of code
calibration to some other materials, such as timber and
masonry. Calibration exercises for these materials have revealed
very high and varied values for safety indexes determined by
current design codes. For example, for timber design, generally
the ratio of the mean to nominal resistance is much higher than
for steel or reinforced concrete members. This difference
appears to arise from the conservatism of timber design strength
rules that must encompass the large variability in material
properties and member imperfections (Ellingwod, 1997).
Traditional stonemasonry skills and knowledge must be
retained and integrated into modern stonemasonry practice
so that a depth and breadth of expertise is maintained. Stone
buildings from past times provide examples of sought-after
skills that are rare, even non-existent today.
Most architectural specifications require that stone meets
certain specified, CEN, ASTM or other testing standards
before it will be accepted for use. For instances in the
European market, the harmonised standards and CE marking
have important consequences for producers. It is the respon-
sibility of the producer, or more correctly the supplier of the
final natural stone product, to ensure that the properties of the
product are documented before the product is sold or being
used in a construction. Since 2006 it has been illegal to sell a
product without such documentation, since the period allowed
for national implementation of the standard is over.
It must be emphasised that the major innovation in the European
standards in contrast to the traditional national standards is that
they are geared to evaluation of conformity and factory
production control. It is now required that the supplier or
manufacturer of dimension stone products shall be responsible
for the assessment and attestation that the products offered are in
conformity with the requirements of European standards. The
compliance with the requirements of the published standards and
stated values, initial type testing and permanent production
control, with registration of result has to be made available.
In Canada and the USA, it is generally required that dimension
stone complies with specified ASTM or other testing standards
before it is accepted for use.
157
Construction Materials
Volume 165 Issue CM3
For example, and relevant here, the determination of flexural
strength under concentrated load or under constant moment
are mandatory and the coefficient of the variation of this
property must be provided for practitioners and designers of
dimension stone. It is quite evident that the time has come to
move forward in step with ‘new’ design and calculation
methods that efficiently and safely meet the needs that must
be overcome by the natural stone sector now and in the future.
REFERENCES
AFNOR (2000) Travaux de bâtiment – Revêtements muraux
attachés en pierre mince. In Partie 1 : cahier des clauses
techniques. CSTB, Paris, France.
ASTM (2010a) Standard guide for selection of dimension stone
for exterior use. ASTM International, West
Conshohocken, PA, USA.
ASTM (2010b) Standard guide for selection, design, and
installation of dimension stone attachment systems. ASTM
International, West Conshohocken, PA, USA.
Beall C (1990) Selecting a Safety Factor for Stone. Aberdeen
Group, Boston, MA, USA.
Bernabéu A and Cura MAGD (2007) The influence of design
requirements on the durability of porous building stones
used in façades. A case study. In: Workshop on Limestone
Decay and Conservation, Malta (Gomez-Heras DM (ed.))
Stone Weathering and Pollution NETwork (SWAPNET).
BSI (1997) BS 8110-1: 1997: Structural use of steelwork in
building. Specification for materials, fabrication and
erection: hot-rolled sections. BSI, London, UK.
BSI (2001) BS 5950-2:2001: Structural use of steelwork in
building. Specification for materials, fabrication and
erection: hot-rolled sections. BSI, London, UK.
BSI (2006) Eurocode basis of structural design BS EN
1990:2002 includes amendment A1:2005. BSI, London,
UK.
BSI (2010) Code of practice for the design and installation of
natural stone cladding and lining. Traditional handset
external cladding. BSI, London, UK.
Camposinhos RS (2009) Revestimentos em Pedra natural com
Fixação Mecânica – Dimensionamento e Projecto. Edições
Sı́labo, Lisboa, Portugal.
Camposinhos RS and Camposinhos RPA (2009) Dimension-stone
cladding design with dowel anchorage. Proceedings of the
Institution of Civil Engineers – Construction Materials
162(3): 95–10.
Camposinhos RDS and Camposinhos RPA (2012) Dimension
stone design – kerf anchorage in limestone and marble.
Proceedings of the Institution of Civil Engineers –
Construction Materials 165(3): 161–175.
CEN (European Committe for Standardization) (2002) Eurocode
1: Actions on structures – Part 1–4: General actions - Wind
actions. CEN TC 250. CEN, Brussels, Belgium.
CEN (2003) EN 13373:2003 Natural stone test
158
Dimension stone design –
partial safety factors: a
reliability based approach
Camposinhos
methods – Determination of geometric characteristics on
units.. Natural stone. Test methods. CEN, Brussels,
Belgium, TC 246.
CEN (2004) Eurocode 8: Design of structures for earthquake
resistance – Part 1: General rules, seismic actions and rules
for buildings. CEN, Brussels, Belgium.
CEN (2006) EN 12372 Determination of flexural strength under
concentrated load. CEN, Brussels, Belgium.
CEN (2008) EN 13161 Natural stone test methods.
Determination of flexural strength under constant moment
Natural stone. Test methods. CEN, Brussels, Belgium.
Conroy K and Hoigard KR (2007) Stiffness considerations in
dimension stone anchorage design. Dimension Stone Use in
Building Construction (Kurt R. Hoigard and Michael J.
Scheffler (eds)). ASTM, West Conshohocken, PA, USA.
Ditlevesen O and Madsen HO (1996) Structural Reliability
Methods. Wiley, Chichester, UK.
Ellingwod BR (1997) Probability-based LRFD for engineered
wood construction. Structural Safety 19(1): 53–65.
Ellingwood AR and Coombs W (1930) The Government and
Railroad Transportation. Ginn, Boston, MA, USA.
Faber MH and Sorensen JD (2003) Reliability based code
calibration – the JCSS approach. Ninth Intemational
Conference on Applications of Statistics and Probability,
2003, San Francisco, CA, USA.
Faber MH, Kübler O and Köhler J (2004) CodeCal. Swiss Federal
Institute of Technology, Institute of Structural
Engineering, Zurich, Switzerland.
Faddy MJ, Wilson RJ and Winter GM (2003) The Determination
of the Design Strength of Granite Used as External Cladding
for Buildings. John Wiley, Sta Lucia, Australia.
Gere AS (1988) Design considerations for using stone veneer on
high-rise buildings. In: New Stone Technology, Design and
Construction for Exterior Wall Systems (Donaldson B
(ed.)). ASTM, Philadelphia, PA, USA.
Grelk B, Christiansen C, Schouenborg B and Malaga K (2007)
Durability of marble cladding – a comprehensive literature
review. In Dimension Stone Use in Building Construction
(Kurt R. Hoigard and Michael J. Scheffler (eds)). ASTM,
West Conshohocken, PA, USA.
Hansen KK, Lekso H and Grelk B (2003) Assessment of the dura-
bility of marble cladding by laboratory exposure compared
to natural exposure. In Research in Building Physics (Jan
Carmeliet HHGV (ed.)). Swets & Zeitlinger, Lisse,
Switzerland, pp. 267–269.
ILI (Indiana Limestone Institute) (1998) Limestone Handbook,
21st edn. Limestone Handbook, Bedford, IN, USA.
Lammert BT and Hoigard KR (2007) Material strength
considerations in dimension stone anchorage design.
Dimension Stone Use in Building Construction (Kurt R.
Hoigard and Michael J. Scheffler (eds)). ASTM, West
Conshohocken, PA, USA.
Lemvig J, Rasmussen J, Jensen JF, Hansen KK and Grelk B (2002)
Construction Materials
Volume 165 Issue CM3
Natural stone panels for building façades – loss of strength
caused by temperature cycles. In Building Physics 2002 –
6th Nordic Symposyum, Trondheim, Norway.
Lewis MD (1995) Modern Stone Cladding: Design and
Installation of Exterior Dimension Stone Systems. ASTM,
Philadelphia, PA, USA.
Lewis MD (2007) Status on development of code requirements
for exterior stone cladding. Dimension Stone Use in
Building Construction (Kurt R. Hoigard and Michael J.
Scheffler (eds)). ASTM, West Conshohocken, PA, USA.
Marble Institute of America (2007) Dimension Stone Design
Manual – VII. Marble Institute of America, Cleveland,
OH, USA.
Melchers RE (1999) Structural Reliability Analysis and
Prediction. John Wiley, Chichester, UK.
Miglio B, Redding J and Al E (1997) Guide to Selection & Testing
of Stone Panels for External Use. CWCT (Centre for
Window & Cladding Technology), University of Bath,
Bath, UK.
Miglio BF, Richardson DM, Yates TS and West D (1999)
Assessment of the durability of porous limestone:
specification and interpretation of test data in UK practice
– stone weathering and durability. In Dimension Stone
Design Cladding: Design, Construction, and Repair
Dimension stone design –
partial safety factors: a
reliability based approach
Camposinhos
(Hoigard KR and Johnson R (eds)). ASTM, Willowbrook,
IL, USA.
Naggatz SG and Gerns EA (2007) Full-scale flexural strength
testing for stone cladding design. Dimension Stone Use in
Building Construction (Kurt R. Hoigard and Michael J.
Scheffler (eds)). ASTM, West Conshohocken, PA,
USA.
Prikryl R and Smith BJ (2007) Building Stone Decay: From
Diagnosis to Conservation. Geological Society of London,
London, UK.
Scavuzzo R and Acri J (2000) In-place load testing of a stone
cladding anchorage system. In ASTM STP 1394,
Dimension Stone Cladding: Design, Construction,
Evaluation, and Repair (Hoigard KR (ed.)). ASTM, West
Conshohocken, PA, USA.
Virlogeux M and Walraven JC (1999) The development of an
international codification for structural concrete with the
CEB-FIP model codes. In Concrete Model Code for Asia,
IABSE Colloquium, Phuket, Thailand, pp. 16–21.
Vrouwenvelder T (2002) Reliability-based code calibration – the
use of the JCSS model code. Joint Committee of Structural
Safety Workshop on Code Calibration, 21– 22 March,
Zurich, Switerland.
Walraven JC (2004) Thinking about codes. Structural Concrete
Journal of the FIB 3: 93–100.
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.
159