Technical ReportPDF Available

TEAM - EU-research project on marble cladding-Final report


Abstract and Figures

Final report of a major European research project on marble cladding: 16 partners and a budget of 4,3 million EUROs. Special focus on durability issues of ventilated marble cladding. The project produced several PhD theses, publications and guidelines. Also standardised tests for evaluation of suitability of marble for this use, one of them subsequently turned into a European standard: EN 16306
Content may be subject to copyright.
CONTRACT N° : G5RD-CT-2000-00233
PROJECT N° : GRD1-1999-10735
TITLE : Testing and Assessment of Marble and Limestone
PROJECT CO-ORDINATOR : SP Swedish National Testing and Research Institute
SP Swedish National Testing and Research Institute SE
Rambøll, Hanneman & Højlund A/S DK
The Foundation for Industrial and Technical Research at the NO
Norwegian University of Technology
Jananders Consulting AB SE
Internazionale Marmi E Macchine Carrara S.p.A. I T
Building Research Establishment Limited UK
Politecnico Di Torino IT
Vienna University of Technology AT
Fischerwerke Artur Fischer GmbH DE
Eka Chemicals AB SE
Nyköping Kommun SE
Georg August Universität DK
Realkredit Danmark DK
Trion Tensid AB SE
Institute of Mechanised Construction and Rock Mining, IMBiGS PL
Slovenian National Building and Civil Engineering Institute, ZAG SI
REPORTING PERIOD : FROM 1 March 2000 TO 31 August 2005
PROJECT START DATE : 1 March 2000 DURATION : 5 Years + 6 months
Date of issue of this report : 31 October 2005
Project funded by the European Community
under the ‘Competitive and Sustainable
Growth’ Programme (1998-2002)
Table of contents
3.1 WP 1 Case studies – Existing buildings 12
3.1.1 Task 1.1 Case Studies 12
3.1.2 Task 1.2 Deterioration mechanisms hypothesis 17
3.2 WP 2 Assessment of facades 23
3.2.1 Task 2.1 Equipment for field measurement of bowing 23
3.2.2 Task 2.2 Inspection of selected Buildings from WP 1 24
3.2.3 Task 2.3 Strength Tests 32
3.2.4 Task 2.4 Risk Analysis 33
3.2.5 Task 2.5 Preparation of Samples 36
3.2.6 Task 2.6 Anchoring System 37
3.3 WP 3 Long term monitoring system 40
3.3.1 Objective 40
3.3.2 Introduction 40
3.3.3 Task 3.1: System installation requirements 40
3.3.4 Task 3.2: On-site monitoring, installation and data collection 41
3.3.5 Data Analysis and Interpretation 43
3.4 WP 4 Selection of samples from quarries and production 46
3.4.1 Objective 46
3.4.1 Introduction 46
3.4.2 Choice of samples 46
3.4.3 Sampling instruction and sampling 46
3.5 WP 5 Research – Finding the mechanism of bowing 50
3.5.1 Objective 50
3.5.2 Task 5.1 Geological Framework (In-Situ Stresses) 50
3.5.3 Task 5.2 Production Conditions 54
3.5.4 Task 5.3 Rock and mineral properties 58
3.5.5 Task 5.4 Other Properties (moisture gradient etc.) 67
3.6 WP 6 Laboratory test methods 78
3.6.1 Objectives 78
3.6.2 Test method for bowing properties 78
3.6.3 Test Method for Irreversible Thermal and Hydric Properties 80
3.6.4 Inter-Comparison trials 86
3.7 WP 7 Impregnation and surface coating 93
3.7.1 Objective 93
3.7.2 Introduction 93
3.7.3 Task 7.1 Development of a laboratory test for bowing 93
3.7.4 Task 7.2 In-situ testing on existing panels 94
3.7.5 Task 7.3 Test field for bowing 95
3.7.6 Task 7.4 Positive side effects 96
3.8 WP 8 Production quality control 98
3.8.1 Objectives 98
3.8.2 Introduction 98
3.8.3 Selection and testing of marbles – Recommendations towards specifiers and suppliers 99
3.8.4 Recommendations to designers 100
3.8.5 Recommendations to suppliers (producers) 101
3.8.6 Elements of importance for FPC 101
3.9 WP 9 Dissemination and exploitation 103
3.9.1 Objectives 103
3.9.2 Activities and results 103
7.1 Introduction/Background 115
7.2 Discussion of the main results 115
7.2.1 Literature study 115
7.2.2 Survey of stone projects 115
7.2.3 Detailed cases studies 116
7.2.4 Long term monitoring 116
7.2.5 Sampling and influencing parameters 117
7.2.6 Full scale laboratory testing, including quarry and processing variables 118
7.2.7 Development of the bow-test and the wet-expansion test 121
7.2.8 Field exposure and possibilities to prevent the bowing or decrease the speed of the ageing 123
7.2.9 Guidelines for production and product control 123
7.2.10 Dissemination activities 124
7.3 Conclusions 124
1 Executive publishable summary
Background: The use of natural stone as facade cladding has been shown to have much
lower life cycle costs, i.e. they are more environmentally friendly than comparable products
of concrete, glass and steel. Promoting the use of natural stone has therefore a great positive
impact on the environment. However, the number of occurrences of bowing and expansion of
marble and limestone panels has lead to increased maintenance costs besides the significant
safety risk and has given negative publicity. The lack of knowledge of a solution to the
problem of bowing marble has a large negative effect on the entire stone trade. In response,
short-sighted and less durable construction solutions are used as alternative, adding to the
decreasing export figures and numbers of employees within the stone sector. The TEAM
project addresses a problem with marble types, from several European countries, that display
bowing on facades in both cold and warm climates. There is, therefore a need to develop
harmonised European standards for differentiating between marble that is susceptible to
bowing and marble that is not. Resolution no. 013, in May 1999 taken by CEN TC 246
Natural Stone states the urgent needs "to develop a direct test method of the bowing risk for
marble cladding products". Thus, the project addresses the mandate for external wall
coverings and the safety of panels.
Objectives: The main objectives were:
To understand and explain the mechanisms of the expansion and loss of strength, probably
the most important phenomena leading to degradation of marble and limestone clad
To prevent the use of deleterious marble and limestone by introducing drafts for European
The project also aimed to develop a concept for assessment of facades, including a
monitoring system in order to predict strength development and improve safety and
To analyse if surface coating and impregnation could prevent or diminish the degradation.
The project has also addressed quality control aspects in order to optimise the production
Work carried out: The TEAM project consortium, representing 9 EU countries, comprised
16 partners representing stone producers and trade associations, testing laboratories,
standardisation and certification bodies, consultants, building owners and care-takers and
producers of fixing and repair systems.
A state-of-the-art report has been written and is based on an extensive compilation of more
than 300 papers on marble and limestone deterioration dating from late 1800s to 2005.
A survey of about 200 buildings has given a clear picture of the extent of the problem in
geographical, geological and climatological terms.
Detailed case studies of 6 buildings have resulted in a methodology for assessment of facades
including monitoring system and risk assessment.
Research both in the laboratory and the field were performed on a large number of different
stone types from different countries and used in different climates. This gave the explanation
of degradation mechanisms and lead to the determination of the critical influencing factors.
Two tests methods, including precision statements: one for bowing and one for thermal and
moisture irreversible expansion have been prepared for submission to CEN TC 246.
Repair techniques based on the use of surface coating and impregnation systems has been
tested at laboratory and in field. Positive side effects including increased durability and easier
cleaning have been observed.
Guidelines for production and product control have been proposed.
An instruction for stone sampling and description has been developed.
Dissemination: The findings have been and will be disseminated mainly through technical
papers, workshops, conferences and the introduction of EN-standards.
2 Objectives of the project
The use of thin marble and limestone slabs as facade cladding has increased substantially
during the last few decades. However, during recent years reports of facade failures have
increased dramatically. Prominent buildings such as the Amoco building in Chicago, the
Finlandia City Hall in Helsinki, La Grande Arch in Paris and IBM Tower in Brussels have all
experienced serious durability problems with their marble clad facades. The problem is
expressed by expansion, bowing, loss of strength, and in most serious cases the detachment
from the anchoring system (Amoco building, USA, Sydsvenska Dagblad, Malmö, Sweden).
The problems regarding limestone facades are slightly different. Apparently, limestone does
not bow, but it expands causing serious problems if the joints are not sufficiently wide to
account for the expansion.
There is a large need of repair systems for existing facades. Despite several European research
projects, the only solution for repair is replacing the panels at large costs of about 400
Euro/m². The recent example of the City Hall in Helsinki, where all panels have been
exchanged in 1998-1999 at a cost of 3.8 Million Euro (the same order of magnitude as the
entire TEAM project) gives an economic perspective of the problem. The problem did not
stop as the new panels were chosen on incorrect grounds. The new panels on the City Hall
started to bow already 6 months after the installation! The today known damaged buildings in
Europe are expected to necessitate repair for well over 240 Million Euro unless other
solutions can be foreseen! Although the vast majority of reported durability problems with
thin marble or limestone slabs refer to the Italian Carrara marble, which is also by far the most
widely used marble in the world. Other marbles e.g. American, Norwegian and Portuguese
have also been reported to bow on facades. However, the reports on performance of Carrara
marble are inconsistent, since in some cases Carrara marble apparently performs satisfactory.
Note that there are about 200 different stone quarries in operation in this area! The direction
of the observed bow may be either convex outward or concave inward relative the facade,
probably mostly depending on environmental (climatic) conditions. However, despite
considerable effort the exact physico-chemical processes responsible for the degradation of
thin marble and limestone slabs exposed to outdoor conditions have not been established by
the research community. As a consequence of the reported durability problems and the lack of
fundamental understanding of the problem, both producers and users (architects and building
owners) of marble and limestone are almost desperate for more knowledge and in particular
they are eager to find a test method able to distinguish durable building stones from non-
durable building stones.
Therefore the project is addressing three clear-cut objectives being:
To establish a sound understanding based on natural sciences of the phenomena leading
to poor field performance of marble clad facades.
To develop a laboratory test method for determination of potential bowing of thin slabs of
natural stone.
To develop a field monitoring, evaluation and repair guide for facade cladding, which
will include risk assessment and service life prediction.
To fulfil the objectives the work aimed:
to update the state of the art.
to establish the service record of a number of marbles and limestones from case studies of
existing buildings in different climatic conditions both of good and poor performance
to select at least 5 buildings for detailed testing and monitoring in order to develop a
monitoring and evaluation program enabling the prediction of the service life of existing
facade claddings
to conduct research studies on vast number of different marbles and limestones in different
environmental conditions leading to comprehensive material characterisation of the materials
including determination of all intrinsic and extrinsic parameters described previously believed
to influence durability as well as a number of other critical parameters
to test different anchoring system and their influence on bowing
to test the repair techniques based on the use of surface coating and impregnation systems
3 Scientific and technical description of the results
Natural stone has been used for facade applications for centuries. Originally, the stone was
rather thick, when used as construction elements, and the durability was apparently good.
Scientific research on properties of marble began in the late 19th century. In the years
following, the thickness of natural facade stones decreased from over 1000 mm (as in
construction elements) to typically 20-50 mm (in cladding applications) as a result of new
cutting technologies and equipment being developed by the industry. Even though most
marble claddings perform satisfactory, durability problems have begun to appear at an
increasing rate after some 50 years of using thin cladding. Well-known buildings such as the
Amoco Building in Chicago, SCOR tower in Paris and the Finlandia Hall in Helsinki (figure
A) have had their marble cladding replaced after less than 30 years at a cost of many millions
of Euros. The deterioration gives a very conspicuous change in the appearance of the panels,
they bow, warp or dish.
Most cases of bowing involve marble from the Carrara area, simply because it is the most
widespread and used marble type. It is, however, vital to emphasize that many building
facades with Carrara marble perform well, and furthermore that marble from other areas all
over the world also exhibits durability problems.
Note that we have coded all rock types of consideration for the quarry owners that have
provided us with test material. E.g. all “Italian marble” come from the Carrara area that has
about 200 quarries in operation, all of them with different properties!
Figure A. Examples of some famous buildings clad with marble or limestone with durability
The bowing of marble is not only restricted to buildings, but gravestones of marble are also
known to bow as is seen in e.g. Montmartre Cemetery (figure B). At present time, most of the
recorded cases are from Europe or North America, most likely because of the much more
widespread use of thin marble claddings in these parts of the world. However, single
recordings have also been made in India (Grabmal des Humayun, Bouineau & Perrier 1995)
and Cuba (Cementary at Havana, Kessler 1919).
Despite more than 100 years of research and being a worldwide problem, the solution to the
problems with bowing marble has not yet been found. Numerous methods have been
employed to investigate possible factors responsible for deterioration of marble. However, no
systematic investigation based upon inspections of building facades with marble cladding
have been carried out before now.
Figure B: Bowing of marble on the Finlandia Hall, Helsinki, Finland (left) and on a
gravestone in the Montmartre cemetery, Paris, France (right).
The holistic approach of the TEAM project takes into consideration the fact that the problem
of bowing and expanding marble and limestone is of interdisciplinary character. This is
illustrated in the work flow scheme below (figure C). The project has therefore engaged
experts from all disciplines concerned, such as stone producers, traders and trade associations,
testing laboratories, authorities/standardisation and certification bodies, consultants, building
owners and care-takers and producers of fixing and repair systems, from 9 European countries
(figure D). The leaders of the different work packages are listed in table A.
WP 1 :
Case Studies - Existing Buildings
WP 2 :
Assessment of Facades
WP 3:
Monitoring System
WP 4:
Selection of Samples from
Quarries and Production
WP 5:
Research WP 6:
Laboratory Test
WP 7:
Impregnation and
Surface Coating
WP 8:
Quality Control
WP 9:
WP 10
Co-ordination and project management
(including interaction with related projects)
Figure C. Flow chart of activities work packages (WP) and their relationships in TEAM.
Figure D. Representation of partners in TEAM.
Table A. Overall responsibilities for the work packages.
WP Activity Leader Country
1 Review of literature and buildings Janaders Consulting SE
2 Detailed investigation of selected buildings Rambøll DK
3 Long-term monitoring of 3 buildings Building Research
Establishment, BRE
4 Sampling of materials UGE & IMM DE & IT
5 Quarry and laboratory testing SINTEF NO
6 Develop of test methods, precision trials SP SE
7 Field exposure sites, remedial activities BRE UK
8 Quality control of products and production SINTEF NO
9 Dissemination IMM IT
10 Administrative and scientific management and
The findings in the Nordtest project "Testing of bowing of marble facade" were of vital
importance for the definition of the workplan in the TEAM project as input, e.g.
for the facade assessment system and the bowing and expansion tests. The general approach
was therefore to start with updating the necessary information from past and present related
projects into a state-of-the-art report of the deterioration mechanisms. In addition, to perform
general case studies in order to gather information about geographically, meteorologically and
geologically differences.
Based on this first step input was mainly given to WP 2 and 3, for the selection of suitable
buildings to be studied in more detail, including the design and installation of long-term
monitoring systems. Monitoring the microclimate in relation to the strain and bowing give
information needed to be able to give recommendations on the possible use in different
climates. It also gives the necessary information to be able to perform a risk assessment, i.e.
to predict further bowing development and decrease in strength (WP 3). Safety criteria can
subsequently be drawn up depending on the type of building, anchoring, locality etc. The
outcome of this second step has been a concept for Assessment of facades. Whether the
anchoring system plays an important role to prevent bowing will be assessed in WP 2.
Additional input to other WP's is the selection of marble and limestone types (WP 4) for
laboratory research of the deterioration mechanisms. In order to get relevant information
about the material properties and performance, a large selection of different marble and
limestone varieties are therefore needed. Among other things, different production
orientations in relation to rock structures in the quarry have been represented among the
samples, and different surface finishing such as grinding and polishing.
The next step, WP 5, was to try to understand the degradation mechanisms by laboratory
research on samples from buildings, quarries and production. Material characterisation was
performed in relation to the findings from WP 1, 2 and 3. Critical levels of temperature and
moisture were determined for the development of test methods. Stress measurements were
carried out in the quarry for the characterisation of the rock mass properties. Critical variables
were then mainly given to WP 6 and 8. As a result of this, one single test method for bowing
and one for irreversible thermal and hydric expansion were chosen within WP 6. The test
procedures have been fixed and the precision established by an inter-comparison test. The
methods and data have been presented to CEN TC 246 Natural Stone.
Previous research has shown that one condition that has to be present for the panels to bow is
moisture. In order to decrease the moisture content and decrease or inhibit further bowing,
different surface coatings and impregnation systems have been tried in-situ and in a field site,
WP 7.
In order to promote a sustainable materials production, WP 8 dealt with the quality control
and how to prevent the production and use of deleterious marble and limestone material.
One of the most important factors for a successful outcome of a project like this is the
possibility to implement the results among the producers, users and authorities. A detailed
dissemination workpackage (WP 9) was therefore described. This WP has included
exhibitions and seminars at annual fairs, Web sites, workshops and a very close co-operation
with the European standardisation of natural stone, CEN TC 246.
Description of research and results in each WP
3.1 WP 1 Case studies – Existing buildings
The work package is divided into the following two tasks:
1. Case studies
2. Deterioration mechanisms hypothesis
3.1.1 Task 1.1 Case Studies
In the TEAM Project Contract Annex A, the following milestones and deliverables were
defined for Work Package (WP) 1 - Task 1.1:
A procedure for inspection of Buildings.
Summary report from Case Studies / Buildings, including selection of potential
Buildings for WP 2 and WP 3.
Selection of potential Buildings for WP 2 and WP 3.
Summary Report on Case Studies.
The following forms and guidelines have been developed
Field Method for Investigation of Bowing of Natural Stone Slabs for Cladding [1]
Form 1.1.1 - Pre Investigation Report [2]
Method Statement – How to approach Building Projects with damaged facades [3]
In addition a database (ORACLE 7.3) has been developed and used for compilation of all data
from investigated buildings.
A total of 194 Building Projects have been identified and reported on various detailing levels
(table 1.1). Among them, 26 Buildings have been selected for further investigation (table 1.2).
All of the 26 Buildings were considered as suitable for dismounting of facade panels
for laboratory testing purposes. Involved partners have tried to get samples
dismounted from the buildings for laboratory testing purposes.
6 Buildings were selected for detailed field studies, measurements of bowing and
supplementary investigations and reports under WP 2 (table 1.2).
2 Buildings were selected for long term monitoring under WP 3 (table 1.3).
We have concluded that the phenomenon of bowing of marble is actually rather common
(figure 1.1). Deformation by bowing is experienced in buildings of various ages, in buildings
exposed to various weather conditions and for slabs of various thickness and dimensions and
with different anchoring methods. Finally, and what is most interesting, bowing is registered
for marble of seemingly very various composition and structure.
Based on our studies we can state that bowing has developed in many more stone varieties
than previously reported in the literature. This has given us an important clue for the further
investigations in the TEAM project.
We have not – as far as investigations in WP 1 is concerned - identified any "artificial
means", used on the market, for preventing, stopping, hindering or delaying the effects of long
term deformation through bowing and subsequent loss of strength. When saying "artificial
means" we refer to choice of stone slabs thickness or dimensions, choice of method for
application (open/closed joints, type of anchoring system, ventilated air slot etc), differences
in climatic regions, applications height over the ground or to various directions of the
Figure 1.1. Yellow boxes mark cities with observations of marble clad buildings (good and
poor). The figure also shows the geographical spread of the problem.
Related to the climate regions/zones it is reasonable also to discuss the effects of different
locations of stone facade slabs on the building itself – height over the ground and directions of
the compass. As for these parameters, our studies are in accordance with the literature:
most pronounced bowing at the upper parts of the buildings and
most pronounced bowing on the building sides facing southeast and southwest.
Pronounced bowing also at facades facing south. The claddings facing to the north show
less bowing tendencies.
This implies that it is the facades receiving the most sunlight that exhibit the highest
percentage of bowing panels and the largest amplitudes.
Further conclusions from WP 1 observations are:
The bowing seems to be related to some types of marble and marble/limestone while other
stone types (travertine, slate, granite, sandstone etc) do not demonstrate this problem.
Both concave and convex bowing ca occur on the same facade with the same marble
(figure 1.2)
There is a clear correlation between bowing behaviour and deterioration leading to loss of
strength (figure 1.3).
It cannot be stated that non-bowing stone slabs are not deteriorating and loosing strength
(e.g. mortised facade slabs seem to be hindered from bowing but might still deteriorate)
There are quite a few different types of marble (with different origin related to quarry
areas) that demonstrate the bowing behaviour.
Marble types or selections from same quarry area or even same quarry can demonstrate
bowing as well as non-bowing behaviour.
Damages and problems of other kind than bowing and deterioration have been registered
and noted but not further discussed within the TEAM programme.
Marble types with known or measured bowing problems and marked strength loss should
be avoided in thin building claddings.
Based on the WP1 observations it is also possible to conclude that revisions in e.g.:
Application system (fixing methods)
Thickness of stone slabs
Panel face dimensions
Placement on the building
may not totally hinder or reduce the deterioration in such marbles.
Table 1.1. Registered Buildings (REGBU) grouped by country location and divided in
"With Bowing Slabs" and “With Non-Bowing Slabs.
Bowing registration acc WP 1 table TOTAL
REGBU Bowing Non-Bowing No Record
Country Quant % Quant % Quant % Quant %
Denmark 37 19 4 11 28 76 5
Sweden 31 16 13 42 18 58 0
Austria 13 7 4 31 8 62 1
Germany 18 9 9 50 8 44 1
Norway 17 9 1 16 0
Portugal 9 5 0 9 0
Slovenia 11 6 5 6 0
Greece 7 <4 0 6 1
Italy 8 <4 3 4 1
Belgium 6 <4 0 0 6
Finland 5 <4 4 0 1
France 5 <4 3 1 1
Spain 6 <4 4 2 0
Croatia 3 <4 1 2 0
Poland 7 <4 2 5 0
UK 3 <4 1 2 0
Switzerland 2 <4 0 0 2
Estonia 1 <4 0 1 0
USA 4 <4 3 1 0
Holland 1 <4 1 0 0
Total Sum 194 58 117 19
Table 1.2. Overview of buildings where a more detailed investigation has been performed
within WP 1.1.
NO Building Resp.
Partner Bow Rock type / Responsible Partner
Collecting Samples
1 SE 02 Malmö, City Hall JAC No Carrara Marble (IT) / JAC
2 SE 03 Gothenburg, School of
JAC No Vratza Limestone (BG) / JAC
3 SE 06A Stockholm, Polstjärnan JAC Yes Bianco Carrara (IT) / JAC
4 SE 07 Västervik, Folkets Hus JAC Yes Bianco Carrara (IT) / JAC
5 SE 14 Borås, City Hall SP No Biacno Carrara (IT)/ SP
6 DK 01 Copenhagen, Realkredit RMB Yes Hove / Porsgrunn (NO) / RMB
7 DK 03 Lyngby, Town Hall RMB No Marmorilik (Greenland) / RMB
8 DK 07 Århus, City Hall RMB No Hove / Porsgrunn (NO) / STF
9 NO 01 Oslo, Dokk-bygget STF No Vencac Beli (YU) / IMM
10 NO 03 Trondheim, Televerket STF No Hove / Porsgrunn (NO) / STF
11 NO 05C Bergen, Berstadbygget STF No Tjøtta (NO) / STF + IMM
12 NO 06 Stavanger, Svanapoteket STF No Porsgrunn (NO) / STF
13 NO 07 Tromsø, Unibygget STF No Norwegian Rose (NO) / STF
14 NO 09 Oslo, Sparebank-1-group STF No Porsgrunn (NO) / STF
15 NO 15 Trondheim, Gamle
STF Yes Bianco Carrara (IT) / STF
16 SF 02 Kouvola Town Hall SP Yes Bianco Carrara (IT) / SP
17 DE 05 Kiel, Bordesholmer Spk UGE Yes Verde Viana (PT) / UGE, JAC
18 IT 01 Torino, Telecom Building PTO No Lasa, Ornovasso (IT) / PTO
19 IT 02 Milano, Gemini Centre PTO Yes Bianco Carrara (IT) / IMM
20 IT 06 Verona, Mazzi Tower PTO No Bianco Carrara (IT) / IMM
Table 1.3. Overview of buildings selected for detailed assessment within WP 2.
No Building Resp.
Partner Bow Rock type WP
1 SE 01 Nyköping, City Hall JAC Yes Bianco Carrara (IT) 2 + 3
2 DK 02 Copenhagen, Danish Nat.
RMB Yes Porsgrunn (NO) 2
3 DE 01 Lünen, Hospital UGE Yes Trigaches (PT) 2
4 DE 03 Göttingen, University
UGE Yes Bianco Carrara (IT) 2 + 3
5 DE 04 Göttingen, University
UGE Yes Peccia (CH) 2
6 IT 07 Magenta Hospital PTO Yes Bianco Carrara (IT) 2
Figure 1.2. A building in Zagreb, Croatia have concave bowing at higher levels of the
building and convex at lower levels at same facade elevations.
Figure 1.3. Severe strength loss associated with bowing.
3.1.2 Task 1.2 Deterioration mechanisms hypothesis Objectives
Preparing a state-of-the-art report based on the latest documentation available and new
interpretations and including the identification of:
1. Marble and limestone types with potential risk of bowing.
2. Deterioration mechanisms.
3. Critical environmental conditions.
The different deterioration mechanism hypothesis shall be listed. The parameters relevant for
performing an adequate inspection shall be pointed out. Based on the results from task 1.1,
previous inspections reports and other relevant literature marble and/or limestone types with
potential risk of bowing shall be identified. Critical environmental conditions for the
deterioration mechanism shall be described and updated State-of-the-art Report shall be
produced. Introduction
Durability of natural stone can be regarded as a measure of its ability to resist decay, i.e. to
maintain its essential and distinctive characteristics of strength and appearance. Durability
may be viewed as the period of time that a stone can maintain its innate characteristics in use.
This period varies with the environment and use, and with the properties of the stone itself.
Durability and deterioration are functions of the intrinsic properties of the rock and the
external environment that are active throughout the lifetime of the natural stone. In order to
understand why some marble types bow and loose their strength when used as thin claddings
in buildings, what reactions that occur between the material and the exterior and how the
decay starts and proceeds, it is necessary to extract information from the literature related to
the different influencing factors. The factors identified include:
Features of the raw material (mineralogy, fabric/texture, pore properties, hydric and
thermal behaviour etc.).
Weather and microclimate.
Construction specific factors. Findings and discussion
It is surprising that, despite several laboratory reports of severe loss of strength, not more field
investigations have been made on this topic. However, investigations have been reported by
Garzonio et al. (1995), Stocksiefen (1996), Jornet (2000), Hook (1994), Wonneberger (1999)
and Mustonen (1993).
Case studies made by TEAM also show large strength loss as shown in figure 1.4; in the order
of 80 % after 35 years for one calcitic marble and 40 % for one dolomitic marble. Laboratory
studies performed by the authors (Schouenborg et al. 2000 and TEAM, 2001b) clearly
indicate that there is no correlation between the amount of bowing and the loss of strength.
This is especially worrying since there is a potential risk of severe strength loss without any
evident bowing of claddings.
0 102030405060708090100
Time of exposure (years)
% of initial strength
Figure 1.4. The relative loss of flexural strength of marble panels exposed to outside climates.
Several hypotheses have been proposed for the observed bowing of marble. Many of them are
contradictory and none fully explains all the different observations and investigations made
by the authors sited in this literature review. Nevertheless, three different directions may be
pointed out:
Anisotropic thermal expansion of calcite and dolomite.
Influence of moisture (and possibly free water) and temperature variations.
Release of locked-in rock stresses.
Erlin (1989, 1999) presents the following 8 steps, which leads to the problems with bowing of
marble panels:
"Thermal expansion of the surface region occurs during high local temperatures.
Individual calcite crystals expand in direction of the "c" axis and shorten in the direction
of the "a" axes.
Calcite crystals, orientated at non-parallel angles to adjacent calcite crystals, "butt" into
the "side" of adjacent.
The contact created forces the dislocation (a phenomenon known as twinning) of the
crystals so that the displaced portion extends into the "space" created by the "a" axis
Shortening of the surface region occurs because of a temperature drop.
The disturbed calcite crystals restrict full return of crystals to their exact original
A volume increase (expansion) occurs in the affected surface region.
Cyclic heating is more dominant on the exposed surface regions than toward the backs of
the panels. Thus there is a volume increase of the face and restraint to that increase by the
rest of the stone. The result is a dished surface.
This hypothesis is however not in agreement with the observation on original cladding of the
Finlandia Hall where the non-exposed surfaces were found to be expanding.
Another hypothesis has been put forward by Winkler (1994). According to Winkler (1994)
the reason for the bowing of marble slabs appears to be a combination of processes that occur
in the following sequence:
"Dissolution triggers micro cracking, followed by dilation from the relief of internal
stresses. Without dissolution the processes of dilation and decay would probably not
Diurnal thermal expansion-contraction cycles continue to dilate the stone, leading to
expansion, bowing, and loss of strength.
Moisture from rain and relative humidity quickly fills pore spaces. Moisture expansion in
heating-cooling cycles dilates the stone further. Bowing of stone panels is normally not
observed in desert areas."
Taking Winkler’s last point one step further it appears that all porous materials have the
potential for "moisture expansion" in heating/cooling cycles. If sufficient pressure is
generated due to thermal expansion of water during heating, all materials will dilate. If wet
marble/limestone is heated it will dilate unless water can escape quickly enough to avoid
pressure generation due to thermal expansion. This theory is in agreement with Svenson
A third hypothesis has been proposed by Garzonio (1995), which includes the following
Built-in strain.
Chemical and physical weathering.
Relaxation and creep stress caused by the weight of material itself.
How the material was laid.
Stress history (applied over a long period of time).
Tectonic-metamorphic processes release of locked-in residual stresses.
However none of the above mentioned hypothesis could explain all the observations. For
example none of the proposed hypotheses explain the apparently beneficial effect of dolomite
content. The "stress history" can be similar for a dolomitic and a calcitic marble but the
tendency for bowing totally different. The problem is that there is no systematic investigation
of rock’s stress magnitude and directions related to the problem of bowing and loss of
strength. Conclusions
Despite the amount of literature available, neither the key influencing factors nor the
mechanisms of the observed deterioration are clear. Several explanations and hypotheses have
been proposed but as yet none explains all the observations in practice. Concerning the loss of
strength it can be concluded that most previous studies have been made on laboratory
samples. However these and observations from buildings (including TEAM) confirm the
seriousness of the situation in that many claddings panels can loose much of their strength
during exposure, regardless of whether they show bowing or not.
As for intrinsic material parameters, most authors suggest that fine grained, calcitic marble
with straight grain boundaries are the carbonate rocks most sensitive to this type of durability
problem. The most important parameters as regards the type of marble appear then to be
composition, grain size and grain interlocking. However, the number of different marble types
investigated by each author is generally very limited and therefore the conclusions concerning
the influence of composition can be incoherent. There is also a lack of agreement concerning
the optimum grain size. A related problem is that the description of grain size is generally
done with one figure, which gives a poor picture of the grain size distribution of the stone.
The inconsistency of micro-structure description also makes it very difficult to interpret the
influence of that parameter. Thus, there is a great need to quantify the micro-structure
parameters and to investigate a larger number of marble varieties.
As for environmental parameters, both temperature variations and moisture are clearly
involved. Moisture has only recently been acknowledged as a key factor. The interaction of
temperature and moisture is believed to be a crucial external factor for the bowing and
strength loss of certain marble types. This is in accordance with both laboratory and field
investigations performed so far within the TEAM project. However, while the moisture in
laboratory specimens can be controlled and define, the moisture contents and gradients in the
construction have not been given much attention in the literature.
As for processing and construction specific parameters, the investigated literature does not
make it possible to conclude about the potential relevance of anchoring system, joint widths,
panel dimension etc.
The risk and rate of degradation of marble cladding increases with:
Increasing temperature and moisture variations and gradients
Increasing idioblastic microstructure
Since it is nearly impossible to restrict the temperature and moisture variations for claddings,
the proper choice of marble type becomes very important. In this respect it appears to be
highly desirable to develop a reliable and conclusive laboratory test method for investigation
of bowing and strength loss sensitivity of carbonate rocks. The need for a reliable test method
can best be illustrated by the fact, that the new marble cladding on Finlandia City Hall started
to bow significantly (figure 1.5) less than one year after replacement in 1999.
Figure 1.5. Finlandia City Hall, Helsinki. The new marble cladding clearly bowed in less than
1 year after the old marble cladding had been replaced. The photo was taken in spring 2001.
NB: Almost all of the old panels were bowing concave, however the new panels bow convex!
In both cases the marble type was a Carrara marble (Bianco Carrara).
Monitoring especially focusing on the influence of the microclimate has been performed (see
WP 3). Studies will also be made on building physics, i.e. ventilation and insulation, in order
to investigate whether it is possible to control the microclimate around the cladding.
It is important to note that the problem of bowing is not restricted to one type of marble or one
climatic zone. The excerpt given below from the published summary of the first years result
in TEAM on this topic states: "In order to gather information and to map the extent of the
problem, about 140 buildings have been selected for classification and/or investigations. The
buildings are situated in Northern, Central and Southern Europe, and there are buildings with
bowed slabs in all countries (Austria, Belgium, Denmark, Finland, France, Germany, Greece,
Italy, Netherlands, Norway, Portugal, Spain, Sweden, Switzerland and United Kingdom)
(figure 1.6). Many of the buildings with problems have been visited and a preliminary
investigation has been conducted. During these investigations several different marble types
from Greece, Italy, Portugal, Spain, Norway and USA have been identified showing
durability problems in terms of bowing."
It is also equally important to draw the attention to the fact that many marble and limestone
claddings and pavements tend to be durable and robust, provided that the correct quality has
been use (e.g. as seen in figure 1.7).
Figure 1.6. Geographical distribution of marble bowing registered by TEAM partners.
Figure 1.7. The Malmö Stadshuset (City hall), S. Sweden; An example of a 30 year old
marble facade with no signs of deterioration or damages.
3.2 WP 2 Assessment of facades
The work package is divided into the following six tasks:
1. Equipment for field measurement of bowing
2. Inspection of selected Buildings from WP 1
3. Strength Tests
4. Risk Analysis
5. Preparation of Samples
6. Anchoring System
The main objectives are to: Develop a method and equipment for measuring in-situ bowing
and expansion of building facades with marble and limestone claddings. Define an inspection
procedure and model for calculation of the safety risk for a facade with marble and limestone
cladding in relation to bowing, reduced strength and anchoring system. Give specific input to
all the involved building owners.
3.2.1 Task 2.1 Equipment for field measurement of bowing Objectives
The objective was to establish a technique and develop a suitable field instrument, a
measuring bridge with a measuring accuracy adequate to measure small continuous changes
from year to year, for in-situ measurement of bowing of building facades.
A bow-meter was developed and tested for precision (mainly reproducibility). The equipment
includes a digital vernier calliper (figure 2.1), readable to 0,01 mm and that can be connected
to a computer for more efficient recording of measurements on building with very high
number of panels.
The reproducibility was measured to better than 0,2 mm. The coefficient of variation is better
than 0,8 %. The equipment is included in the method for in situ investigation of facades [4].
Figure 2.1. The "Bow-meter" has been produced by gluing 3 aluminium profiles together to
increase the strength and decrease the weight.
It is always crucial to measure the magnitude of the bow versus a flat reference surface. Since
is not always possible to find a flat surface, a zero-set device was developed (figure 2.2).
Figure 2.2. Precision gauge block for zero-setting of the Bow-meter, equivalent to a
reference plane.
3.2.2 Task 2.2 Inspection of selected Buildings from WP 1 Objectives
A detailed inspection of 6 selected buildings from WP 1 has been planned to be carried out.
The task includes documentation of the amount of bowing using the measuring device from
Task 2.1, selection of samples of slabs, reports on the anchoring systems, and a more detailed
description of any damage effects and signs of deterioration. An increased number of cladding
panels has been measured in detail, compared to WP 1, to ensure a statistically sound basis for
assessment of the risk analysis. Introduction
A number of buildings with marble cladding has been reported to have deterioration problems
with bowing and expansion of the panels, leading to loss of strength, cracking and break-outs
of fixing systems and thus to potential safety problems with a risk of panels falling from the
A part of the TEAM project has therefore dealt with the detailed inspection and assessment of
the buildings. A total of 6 buildings with bowing facades have therefore been subjected to a
detailed inspection and assessment as test cases in order to design suitable inspection and
assessment procedures.
Based on the investigations on the 6 selected buildings it is possible to point out some
valuable inspection methods and results for each of the objectives. The observations (table
2.1) and the conclusions which are relevant for the inspection and assessment of the bowing
and expansion phenomena are presented in the following sections. The inspection results are
in this report not compared to the later results of the laboratory testing of the sampled panels,
but are discussed in other reports. Age of the panels
Bowing and expansion will increase with the age of the panels, and it is preliminarily
assumed that these will develop, proportionally to the exposure time. The inspected buildings
and the literature show that already after a few years of exposure (< 5-10 years), bowing and
deterioration may occur in panels.
Knowledge gathered from the buildings outside the TEAM project supports actually the
assumptions that bowing (and strength loss) is linear correlated with the age and time of
exposure. However, the observed rate of development varies from one panel to another, even
in the same building. This rate can to some extend be determined by measuring bowing and
joint width for a larger number of panels.
Repeated measurements of e.g. joint width or bowing will over a longer time provide
documentation for the expansion rate in the actual building. This may be obtained by
permanent, long-term monitoring or by repeated inspection with 2-5 years intervals using bow
meter. The concept of measuring the joint widths for a range of panels with 2-5 years interval
was tried at the Danish National Bank in 2001 and 2004 where the additional 3 years
exposure corresponded to 10-15 % increase of the exposure time, but this indicated no
decrease of the joint width over this period. Panel dimensions
Bowing is observed for all panel dimensions (largest dimension ranges from 900 to 2000 mm)
and all thickness ranging from 30 to 60 mm. In the literature some scientist mentioned an
optimal thickness due to bowing but so far it has not been possible to verify this during the
The inspections and the literature shows clearly that larger panel dimension (width and
height) will increase the bowing, since the bowing amplitude (U) is correlated with the length
(width) (l) and the thickness (t) of a panel. The physical equation should be obtained:
U = Δε l2 / 8 t, where Δε is the difference in strain from the front to the rear side of the panel.
The loss of strength of smaller panels may therefore be equally severe, even if the smaller
panels show smaller visible bowing amplitude, just as expanding panel with no visible
bowing may loose strength.
The bowing and expansion ultimately lead to increased forces at the fixing points or lines and
the dimensions can therefore often influence the resulting forces, which need to be transferred
and thus to the increased risk of failures.
Table 2.1. Summary of information on assessed buildings.
27 Facade orientation
The bowing observations indicated significant differences in the bowing amplitudes
depending on the facade orientation (figure 2.3). It appears from the investigations that the
bowing phenomenon is less significant to facades not or less exposed to temperature
differences during a day.
It is experienced that the largest bowing amplitudes are usually not found on the north
facades, because these are not exposed to large temperature cycles during a day.
For several buildings or part of these more pronounced bowing was observed on the east or
west facades than on the south facades. A reason for that could be that a low sun angle might
course larger temperature variations during a day than a high continuous sun angle. That
might have influence on the bowing behaviour between different facade orientations on a
certain building. Besides this, it is a fact that the sun angle varies from north / south to the
equatorial zone, and that might give some explanation why bowing behaviour differs between
different zones.
Figure 2.3. Bow measurements at Danish National Bank in 2001, 2nd row from the ground.
Scale for bowing is in steps of 1 mm/m.
28 Height above the ground
For the buildings in Göttingen (see figure 2.4) and the building in Nyköping it is observed
that the bowing phenomenon increases with the height above ground. This may be explained
as the exposure and the variation of the climate (wind, humidity, temperature) increases with
the height above ground.
Figure 2.4. Building map of the southern facade of the Theologicum Building (true to scale).
Different colours represent panels with different bowing amplitudes (colour code see figure
Convex bowing
Concave bowing
Figure 2.5. Definition of bowing value, deformation classes and colour codes. Climatic conditions
From most of the investigated buildings it is observed that the absolute temperature and the
temperature variations seem to be very important factors for bowing to occur. The higher the
temperature variations the higher the degree of bowing will be. Varying temperature gradients
through the panel thickness may have an influence on bowing, but this has not been verified
by the inspections.
On the Danish National Bank and other structures larger temperature variations were
measured on the front side than on the rear side, since the back side is sheltered from direct
rain and sun exposure. This may have influence on the degree of bowing and also the bowing
direction, but the influence can not be verified by the observations so far. The humidity might
also be an important factor due to bowing, but based on the building investigations it is not
possible to draw any conclusion. However, it is observed in the laboratory that bowing only
occurs when free water is available. A dry screening test results in expansion only and no
bowing. Fixing/anchoring systems
Bowing, concave and convex as well, is observed for all fixing systems represented in the
investigated buildings. Based on the investigations no specific fixing system can be pointed
out as being more (or less) suitable to prevent bowing, expansion or the development of
The type of fixing system, influences the bowing shape at a late stage of the bowing (sever
bowing) and the type influences significantly the type of cracking and spalling observed in the
buildings. Cracks and breakouts
Cracks and stone damages are observed near the supporting dowel anchors or kerfs. An
increase in bowing is correlated with an increase in the relative frequency of damages e.g.
cracks and breakouts (figures 2.6, 2.7 and 2.8).
Figure 2.6. Typical crack out of fixing pins at Göttingen university library.
Figure 2.7. Panel fixed by fixing pins at top and bottom, cracked by bowing.
Figure 2.8. Cracking at corner, typical for panel anchored by rail at Theologicum. Cladding design
Bowing, concave and convex as well, is observed for all cladding design represented in the
investigated buildings. Based on the investigations no specific conclusion can be drawn. Open or closed joints
The use of open joints may lead to larger variation in the environment behind the panel, in
contact to the design with closed joint, where the air behind the panels may be sealed in and
isolated from the outdoors. No effect of this has been verified by the inspections, however,
facade claddings with elastic sealant in the joint will often reveal if an expansion has taken
place (see figure 2.9).
31 Width of the joints
For most of the buildings it is observed that the average widths of joints are below the design
width. If a panel is strained when free movement in the joints is not allowed, normal forces
will occur in the panel or it may lead to a pushing aside of the neighbouring panels. The
restraining may cause the shape of the bowing panel to change, e.g. to bowing predominantly
in one direction and with damages typical near the fixings.
On Danish National Bank the investigations indicate a linear correlation between the absolute
bowing amplitude and the mean value of the joint width. The smallest joint width is found
where the bowing is strongest. It is assumed that this is a fact for all investigated buildings.
Figure 2.9. Elastic joint between expanded panels. Notice the squeezed out elastic sealant. Fabric/orientation of foliation
On most of the investigated buildings the orientation of foliation could not be observed but on
one building (Oeconomicum in Göttingen) it was possible to investigate that the cutting
direction, with respect to the metamorphic layering, foliation or macroscopic folds had
influence on the bowing amplitudes and shapes. Convex or concave bowing
It has not been possible to explain why panels are bowing in the concave or convex shape on
a building.
One theory was that panels exposed in the same way for example on a part of certain facade
would bow in the same direction, but it has repeatedly been observed that some few panels
bow opposite than the rest. It has in addition to this been observed that new fresh panels
mounted instead of demounted panels are bowing opposite the demounted panels. Surface finishing and treatment
After the assessments it has been discussed whether the surface finishing and/or the cutting
processes may influence the bowing and the start conditions in the beginning of a panel’s
lifetime. Possible surface finishing could be mechanical (rough, smooth or honed) or
treatments like chemical (impregnation, soap etc.) and the cutting process could be more or
less "rough".
No conclusion concerning finishing could be reached based on the results from the 6 building
inspections. Sampling and strength tests
The visual inspection and the measurements of bowing and joint widths provided a good
overview of the conditions. However, the detailed assessments clearly showed that the effects
of the bowing and expansion could only be determined by taking samples form the exposed
and deformed panels in combination with "fresh panels". This should normally include:
One panel, from the most exposed part of the structure, where most bowing has been
One panel, from a moderately exposed are, where the panels bow moderately.
One panel, from the least exposed part of the structure, where no or almost no bowing
has been observed. This panel shall preferably be taken from the reserve panels (a
building owner will often have a reserve of panels intended for replacement of cracked
panels and such panels have never been exposed to the environment).
These panels are normally used for determining the actual loss of strength and for carrying out
laboratory testing for bowing and expansion and additional strength test, so the observed
bowing can be roughly correlated to strength loss. Short conclusion
The 6 detailed assessments have provided a lot of general knowledge of the deterioration.
Most important is the establishment of a basis for the development of inspection guidelines
and procedures for inspection, assessment and sampling for the later risk based assessment of
the facade cladding. The guidelines are provided as a separate report.
3.2.3 Task 2.3 Strength Tests Objective
The objective was to analyse the relation between the bowing, expansion and ageing of
marble and limestone slabs received from Task 2.5 and a reduced strength if any of the same
slabs. Strength testing includes as a minimum following tests: flexural strength and break load
of dowel holes.
This task was later integrated in WP 5 laboratory testing, including materials from quarries,
production and buildings. No further comments will therefore be given here.
3.2.4 Task 2.4 Risk Analysis Objectives
Set up an evaluation and prediction model for risk analysis of marble and limestone cladding
failure and carry out the risk analysis for the buildings inspected in Task 2.2. The model is not
foreseen to be computer based. The model shall be submitted after 24 months to WP 3. Introduction
Although a considerable amount of information about the performance of marbles has been
collected from previous investigations and research projects the data is still far from complete
and so it is not possible at present to build a general predictive model to calculate the safety
risk for all kinds of marble and limestone cladding. The reason for this is that the data tends to
be very specific to particular sites, environments, and stone types. Not enough data has been
collected yet to establish any correlation of the factors that influence the risk of failure
between the sites reported on.
The key factors in determining the propensity to bowing, loss of strength and facade
performance are:
Marble type
Panel size and thickness
Material Strength
Prevailing weather – wind loading, temperature & moisture
Assuming the observed and predicted decay rates to be reasonably correct it is possible to
calculate the time for the strengths to fall below the design wind loading and the time for the
strengths to fall below the limit where the factors of are unity.
The point at which the design strength and panel strength match in merely informative as at
this point the factors of safety are at 8 and 6 respectively for flexure and shear. These factors
are present to accommodate such issues as variability in the stone, workmanship, material
decay and other uncertainties.
Obviously there is cause for concern where these factors of safety have been substantially
eroded – hence giving an estimate of the possible time scale is important.
Applying all of this data allows an estimate of risk to be made where risk is the product of the
probability for "failure" and the consequences of that failure. For many facades the
consequences of a panel falling into the street are very great and so even a very small
probability of this occurring gives rise to an unacceptable risk. Description of a risk analysis model for marble
Although a considerable amount of information about the performance of marbles has been
collected from previous investigations and research projects the data is still far from complete.
It is not possible at present to build a general predictive model to calculate the safety risk for
all kinds of marble and limestone cladding. The reason for this is that the data tends to be very
specific to particular sites, environments, and stone types. Not enough data has been collected
yet to establish any correlation of the factors that influence the risk of failure between the sites
reported on.
There is no single theoretical basis for the various modes of failure reported on. However,
there a few common factors linking some of the failures that can be used as a starting point
for a risk analysis. These are:- temperature cycling caused by sunshine, mineral "fabric",
fixing method, and cladding design. As more data becomes available from case studies a
much better correlation among the modes of failure and deleterious effects should emerge.
Ultimately, failure mechanisms and theories should enable quantitative predictions to be
The approach that has been taken is to establish a framework that will be populated as data is
collected during the project. An active document scheme has been started that puts current
knowledge into a structured form that can be easily browsed in a similar way to Internet web
browsers. The user is guided through the document to the information relevant to their task.
The information in the document is in the form of text, tables and graphics but other media
and methods can be added such as sound and calculations or any other suitable for Internet
browsers. The document can have forms that can be filled in to direct the user to the
appropriate parts of the document. As more data becomes available from other parts of the
TEAM-project, external commissions etc, sections of the hypertext document can be made
more specific. Where formulae and test methods are available it will be possible to perform
calculations and make specific recommendations. The proposed model
In its present form the model can be viewed with the current Windows™ help system. This is
based on compiled HTML (an Internet web page format) documents. Requirements for using
the system are Windows 9x or later with the Internet Explorer version 4 or later also installed.
The Windows™ help system displays the information in a similar way to a web browser (such
as the Microsoft Internet Explorer) but with some automatic features that provide in-built
indexing and search capabilities. These features can be made accessible for expert users
wishing to access specific areas of information. Non-expert users will not necessarily have an
index available but will be guided through the documents depending on choices they make as
they progress.
The textual basis of the HTML documents is compiled from a Microsoft Word™ document
that has been extracted from the State-of-the-art report. The extraction is not yet complete.
Information from the case studies will be included as it becomes available.
The document is structured under the headings as outlined below:
Environmental Factors
Air pollution, acid gasses etc.,
Freezing and thawing cycles
Orientation to the sun and prevailing weather
Construction Specific parameters
Anchoring system
Surface treatment
Age of facade
Production and processing techniques
Geological framework
Mineralogy – primarily calcite/dolomite
Grain size, grain interlocking, grain boundaries
Preferred orientation of crystals
Porosity, pore size distribution, permeability, water absorption
Physical (mechanical) Properties
When viewed from within the Windows™ help system the user can select the next topic by
"clicking" and appropriate button. A screen shot of the introductory page is shown below
(figure 2.10).
Figure 2.10. Example of an active document format.
As the scheme is developed the user will arrive at guidance relevant to their task. Where
possible at least some "boundary conditions" will set, using current data from the case studies,
to establish the high/low, best/worst states.
A more specific evaluation of service life and risk will be gained as the findings and ideas
from the case studies are linked with loss of strength and the fixing and wind loads.
3.2.5 Task 2.5 Preparation of Samples Objective
Development of an instruction for sampling and preparation of samples received from Task
2.2 for laboratory testing of strength, bowing and petrography.
The document WP 2.5-SP-TD-020621-Samples from buildings-rev 4 gives guidance of how
to sample and prepare samples for testing in order to be able to evaluate the material
properties and get a general idea about the present strength and strength development over
time for the actual building.
Three principally different types of panels shall be sampled whenever possible:
Fresh material from spare material or inside mounted
Strongly/clearly bowed slab from a very exposed placed
Slab from the outside but taken from a sheltered place
Normally five to six panels will be sufficient to get enough specimens for every test and get a
spread in the strength properties represented among the samples.
Test specimens shall be cut in order to represent any directional dependency, inner and outer
(the frame) parts of the panel. One example of cutting pattern is illustrated in figure 2.11
In addition to traditional testing of physical properties, it is also proposed to test bowing
tendencies and to use non-destructive (NDT) testing like ultrasonic pulse velocity (UPV). If a
relationship can be established between strength and UPV, it is also possible to monitor the
changes of the facade at repeated visits without demounting any more panels.
Figure 2.11. The photo illustrates where test specimens will be cut.
3.2.6 Task 2.6 Anchoring System Objective
Investigation of the influence of the anchoring system on the bowing and expansion of marble
and limestone panels based on input from Task 2.2. The taskleader (Fischerwerke), together
with Jananders Consulting, carried out the investigation and make a report on the use of
various anchoring systems and describe if possible whether they have been designed to
account for any anticipated bowing. Summary of the survey
The TEAM inspections of buildings (about 200 projects in Europe) related to different
Anchoring Systems used for thin Stone veneers– combined with knowledge from previous
experience - have resulted in a listing of Application types as follows:
1. Mortised with or without tying back fixings at edges or back faces.
2. "Traditional" anchors with dowels at stone edge dowel holes.
3. Brackets or rails at horizontal Stone edge kerfs.
4. Full floor level high facade sections resting on the bottom Stone edge.
5. Back face fixings on special sub-frame systems.
6. Prefabricated concrete units with Stone panels tightly attached to a concrete layer.
7. Simple sandwich units composed of a Stone panel attached to a thin concrete layer.
8. Honeycomb supported Stone panels installed by special sub-frame systems.
Type 1) installations have been observed at a limited number of building projects. It has been
noted that marble types that deform on other types of installation have not been observed to
do so when mortised.
Type 2) is the most frequently used version of application principles of the inspected
buildings. The positioning of anchors defines a reference plane. In cases where the marble
panels have deformed by curving or dishing the anchors do not have any influence on this
phenomenon, other than the shape of the bowing. The anchor location defines a permanent
reference plane and the Stone panel deforms related to this plane.
Type 3) is registered at two projects (Göttingen Theologicum and Magenta Hospital, Torino)
with a deforming type of marble. The rails define a reference plane around which the marble
panels deform. This type of anchoring is the most common one in the USA.
Type 4) was not registered among inspected buildings within TEAM.
Type 5) is frequently used but there is no knowledge of projects with marble or limestone.
Opposing to the other types of installation this type is characterised by the fact the Stone
panels are hanging from the fixing points while the Stone panels installed by the other types
of application are resting on the fixings – or in some few cases carried over the whole surface.
Type 6) have a similar principle structure as Type 1) but the building projects with this type of
built up were not covered with marble. They were covered with other types of Stone known
for having no tendencies for bowing or degradation.
Type 7) was not registered among inspected buildings within TEAM.
Type 8) was not registered among inspected buildings within TEAM.
38 The Fischerwerke test wall
The partner Fischerwerke has installed a test wall (figure 2.12) and performed measurements
continuously once a month starting in July 2003. Readings have been reported in a formula
prepared for its purpose. Selected parameters have been collected and summarised in
Mortised Continuous Backface Dowels at Dowels at
Rails FZP fixing vertical horizontal
joints joints
1 4 7 10 13
2 5 8 11 14
3 15 6 9 12 Conclusions
Three main conclusions can be drawn from this study:
1. Panels installed mortised are bowing significantly less then the other types. Panels
installed on continuous rails have the most significant bowing and the three remaining
types show roughly identical bowing behaviour (figure 2.13).
2. All slabs located in bottom row bow less then all slabs in top row. Slabs in middle row in
most cases have shown amplitudes between those of the top row and the bottom row. See
curve diagrams figure 2.14 below.
3. The type of application has an obvious impact on the extent of bowing but is not a method
to avoid bowing. Pending on the "rigidity" of the application, the bowing will be more or
less significant.
Figure 2.12. Image of test wall for anchoring systems completed
with marble panels slab numbers and type of fixing system
lied at each set of columns.
Fischerwerkes Test Wall at Waldachtal
2003-06-28 2003-10-06 2004-01-14 2004-04-23 2004-08-01 2004-11-09 2005-02-17 2005-05-28
Bowing (mm)
Figure 2.14. Bowing amplitudes – Dowel vertically. Wall slabs bowing development for top,
middle and bottom rows.
Fischerwerke's testwall
03-06-28 03-10-06 04-01-14 04-04-23 04-08-01 04-11-09 05-02-17 05-05-28
Bowing (mm)
Figure 2.13. Mean values for bowing (always convex, i.e. bowing outwards) for
each type of application.
3.3 WP 3 Long term monitoring system
3.3.1 Objective
The principle objective of this Work Package was to develop and evaluate a system to
continuously monitor deformation/movements for panels and to monitor environmental
conditions in order to determine if bowing is occurring and the changes induced with time.
The task within the WP included developing a procedure and requirements for long term
monitoring and to provide input to the calibration of the risk model in WP2.4.
3.3.2 Introduction
This work package has two main monitoring parts, to monitor deformation/movement for
panels and to monitor environmental conditions. The aim was that all tests and measurements
performed would be non-destructive.
The task necessitated the development of equipment for long term monitoring, installation of
equipment at three locations and processing of the results. The data has been used to calibrate
a model of the risks of failure associated with the changes in the properties of the panels over
long periods of time in order that these risks can be better understood and managed.
3.3.3 Task 3.1: System installation requirements
This task was used to choose suitable equipment and monitoring systems - temperature/
humidity sensors, strain gauges, data logging – and to select suitable buildings – with access
to front and rear of panels. This task, together with Task 3.2, were used to determine the
necessary data logging frequency and protocol for collection and analysis of data. It was
decided that we would start with an easily accessible location to check that the system works
The selection of parameters to be measured, selection of equipment and selection of suitable
buildings were all given careful consideration before being agreed. It was agreed that ideally
at each location monitoring would include:
surface temperature on the external surface of the stone
strain in two directions on the external surface of the stone
surface temperature on the internal surface of the stone
air temperature in the gap behind the panel
relative humidity in the gap behind the panel
strain in two directions on the internal surface of the stone
shade air temperature
shade relative humidity
There were problems with finding ideal sites and so there was a need to ‘compromise’. The
final list of sites is:
Danish National Bank, Denmark
University Library Göttingen, Germany
Nyköping City Hall, Sweden
It was planned that additional monitoring may be undertaken at some of the other buildings in
the project but that this monitoring would be short term and focussed on specific questions.
3.3.4 Task 3.2: On-site monitoring, installation and data collection
The first two sets of equipment were installed in Copenhagen at the beginning of September
2001, two sets of equipment were installed in Göttingen in November 2001, and three sets at
Nyköping in October 2002 (figures 3.1 and 3.2). The ‘sophistication’ of the equipment has
progressed as the team has gained further experience of the sites and has overcome the initial
problems relating, for example, to the fixing of the strain gauges. The three sites are described
Figure 3.1. Photo of the three panels monitored at Nyköping City Hall.
Danish National Bank
Amongst the most interesting results from the Danish National Bank in Copenhagen is the
fact that there is very little difference between the temperatures on the front and back of the
panels – only a few minutes ‘lag’ (figure 3.3). On the eastern side the front can be 4 ºC
warmer when the sun first shines on this face. The full temperature rise can take 6 hours but in
more extreme cases the rate is around 0,3 °C per minute and this information has been used in
the development of the bowing test.
Figure 3.2. Solar cell charged logging equipment at Nyköping City Hall
Göttingen University Library
The equipment has been in place since November 2001. The aim of the logging at this site
was to evaluate a new type of temperature and humidity logger and to evaluate the first strain
gauges for their thermal stability and the magnitude of movement.
Most of the equipment seemed to work well but there were concerns about the strain gauges
as the data shows sudden changes in strain. Despite some problems with the downloading of
data interesting results have been found with diurnal cycles clearly visible against a
background of longer term changes with reliable stable data being obtained over short periods
– for example 7- 8 days.
However, over longer periods the ‘drift’ and background strain readings still seem to show
sudden changes.
Figure 3.3. Temperature and humidity data, Copenhagen
Nyköping City Hall
The equipment was installed on the City Hall at Nyköping in October 2002 – this was timed
to coincide with the setting up of the field exposure site. The aims at this site are to evaluate
the re-designed strain gauges, in particular to look at the magnitude of movement, expansion
vs bowing and the effects of surface treatments.
The strain gauges are a new design using ‘Invar’ to reduce the thermal expansion of the
gauges and provide more stability in the results. There are three monitoring locations – one
location is on an existing Carrara panel (that has been demounted and replaced), the second is
on a new panel of Carrara with similar dimensions to the existing panel. The third location is
on a panel that has been treated with a microcrystalline wax from Trion Tensid. In all three
cases, strain gauges have been fixed to the rear and front faces to allow determination of
movement in three dimensions. Further analysis allows this to be converted to a ‘bowing’
figure which can be compared directly to the ‘bow meter’ measurements being made on site.
All of the data is collected by a single data logger. Only one set of temperature and humidity
measurements is being made behind the panels and in the air as experience from Göttingen
has shown that there is very little variation between adjacent panels.
3.3.5 Data Analysis and Interpretation
The diurnal movement in the strain gauges can be clearly registered with this equipment and it
is also clear that this daily movement is likely to ‘mask’ any longer term residual movement
in the panels from the early expansion and/or bowing of the marble. As a result a considerable
effort has gone into the development of methods to ‘process’ the data. Three approaches have
been used:
Selecting and examining strain gauge data recorded over a narrow range of
Selecting strain gauge data recorded at the same time each day
Correcting the entire dataset (14,000 points) for thermal movement of the strain
Figures 3.4 -3.6 show the results from applying the first method. The residual bowing of the
panels can be seen over the 15 months of the monitoring.
Figure 3.4 shows Panel 1 which was removed from the building in November 2002 and
replaced on new fixings. The pattern of movement seems to show a fairly rapid bowing
immediately after the panel was replaced – possibly as it moved back against the fixings-
followed by a long period with very little movement.
Figure 3.5 shows Panel 3 which was a panel of new Carrara marble. Note that the panel is
bowing in the opposite direction to Panel 1 and that the magnitude of the bowing is also much
Figure 3.6 shows Panel 2 at Nyköping which was a panel of new Carrara marble that was
treated with a hydrophobic coating. The treatment was intended to reduce the bowing. Initial
the rate seemed to be reduced but it is not clear now quite what is happening except the panel
seems to be move far more on the horizontal axis than on the vertical axis.
Calculated gap under a 1m straight-edge
0 50 100 150 200 250 300 350 400 450 500 550 600 650 700 750 800 850 900 950 1000 1050
Gap (mm)
7.3 ± 2 °C Panel 1 H Gap
7.3 ± 2 °C Panel 1 V Gap
Data collected between October
2002 to May 2005
Oct - Dec
ril - June Jul
- Se
tOct - Dec
ril - June Jul
- Se
tOct - Dec
ril - JuneJan-MarJan-Ma
Figure 3.4. Panel 1 at Nyköping – an existing panel of Carrara Marble removed and refixed.
Calculated gap under a 1m straight-edge
0 50 100 150 200 250 300 350 400 450 500 550 600 650 700 750 800 850 900 950 1000 1050
Gap (mm)
7.3 ± 2 °C Panel 3 H Gap
7.3 ± 2 °C Panel 3 V Gap
Data collected between October
2002 to May 2005
Oct - Dec
ril - June Jul
- Se
tOct - Dec
ril - June Jul
- Se
tOct - Dec
ril - JuneJan-Ma
Figure 3.5. Panel 3 at Nyköping – a new panel of Carrara Marble
Calculated gap under a 1m straight-edge
0 50 100 150 200 250 300 350 400 450 500 550 600 650 700 750 800 850 900 950 1000 1050
Gap (mm)
7.3 ± 2 °C Panel 2 H Gap
7.3 ± 2 °C Panel 2 V Gap
Data collected between October
2002 to May 2005
Oct - Dec
ril - June Jul
- Se
tOct - Dec
ril - June Jul
- Se
tOct - Dec
ril - JuneJan-Ma
Figure 3.6. Panel 2 at Nyköping – a new panel of Carrara Marble treated with a hydrophobic
3.3.6 Discussion and findings
The key finding since the initial installation requirements were determined is that there are no
"off the shelf" systems available for this type of monitoring. In particular, all the strain gauge
systems required installation conditions which cannot be achieved on existing marble panels -
for example a polished and clean surface or heating to 70 ºC to set the adhesive for the
gauges. As a result considerable effort has been put into the development of a system which
is suitable for measurement on an existing building. Evaluation of the initial observations
from Copenhagen and Göttingen have been used to design more stable equipment for the
measurements at Nyköping and this new equipment seems to be working successfully. The
earlier work at the Danish National Bank has shown that it is possible to reduce the data
logging frequency with no loss of precision in the results and the analysis of the data from
Nyköping has show that it is possible to determine the bowing of the panels by filtering the
data to specific temperatures and so removing the need to correct all the data points for
thermal movement of the gauges.
Comparison of the strain gauge monitor and the bow meter measurements shows the results to
be very similar which indicates that the "real time" strain gauge measurements can be used to
supplement and strengthen the findings from the bow meter measurements.
The methods to "filter" diurnal changes have provided important input to the interpretation of
data from field exposure on Nyköping City Hall and at the various field sites established at
the TEAM partners.
3.4 WP 4 Selection of samples from quarries and production
3.4.1 Objective
The purpose of the task has been to select different marble and limestone types mainly on the
basis of the results from WP 1, 2 and 5 (screening test), taking into account how they are
quarried and further processed in respect of different tectonic directions in the rock mass. By
this is meant that e.g. important anisotropy planes, cleavage directions etc. shall follow the
rock material into the final specimen preparation. The work has implied the following
Decide about rock types and varieties that shall be selected
Describe the stone types and the quarries using the forms from Task 4.1
Agree with other WPs about number, cutting directions and sizes of samples and
specimens needed for laboratory tests in WP 5, 6, and 7
Perform the sampling and provide samples for other WPs
Use and update the data compilation system from Task 4.1
3.4.1 Introduction
The main work has focussed on trying to link observations made in WP 1 and 2 to the
selection of samples from quarries, to define crucial parameters of the stone, discuss rock
mass properties, incl. rock stresses and the quarrying and processing that may influence the
sampling strategy. (WP4.2-UGE-TD-010425-Selection of samples from quarries and
production-State-of-the-art and WP4.2-STF-TD-050930-State-of-the-art-Part-Rock Stresses).
Finally, to define detailed instructions for the sampling and sample identification to assure
traceability throughout the project [5].
3.4.2 Choice of samples
The information about the marble used for any building is very poor. In general, the type of
marble is known. The quarry is not known. Considering the many individual quarries in a
quarry district it is easy to understand that the possibility to select the exact same material that
is represented on a building is close to zero. The decision was therefore taken to try to trace
the material as far as possible and then choose the most similar marble to the ones present on
the buildings investigated in WP 2.
In total, 16 quarries have been sampled, giving 17 varieties of carbonate rocks (table 4.1). The
sampled rock types represent 3 pure dolomite marbles, 11 pure calcite marbles, 1 ophicalcitic
marble (containing serpentine), 1 limestone and 1 silicate rich, contact methamorphic
limestone. In addition a large number of marble types were chosen for test screening for later
selection of additional stone types to include in the comprehensive testing scheme of WP 5
(table 4.2).
3.4.3 Sampling instruction and sampling
A "Rock Assessment form" [6] has been developed and used (included in the report WP4.1-
STF-IMM-UGE-TD-011128-Sampling method statement Quarries) for the description of the
sampled stone types as well as the quarries from where the specimens were selected. For the
purpose of TEAM, it has not been a goal to map or to be in full control of quality variations in
each quarry, but to select one variety (one block) and assure that all laboratories and partners
perform the analyses on the same block material and on specimens cut in the same way.
The importance of traceability and to be present and supervise the sampling has been
emphasised during sampling in WP 4. Therefore, most of the blocks were sampled and
marked under control of at least one geologist and member of the TEAM research group.
In general, one block of sufficient size was selected from an existing quarry face, or
alternatively, a block was selected from a block storage or processing plant. The sampling in
Slovenia and Poland was done in 2003, the rest in 2001-2002. The blocks chosen for cutting
into test specimens were oriented in case of visually present fabric (figure 4.1). Drill cores
were also taken in order to try to detect the presence of any preferred orientation of minerals
(figure 4.2).
Figure 4.1. Principal sketch of cutting orientation in relation to the fabric of one of the stone
In order to secure confidence, a coding system was established during sampling and marking
(table 4.1). The first two letters design country of the origin, "q" stands for quarry (in order to
separate these samples from samples from buildings) and numbers from 1 to 4 for the number
of the quarry or the variety, respectively (e.g. chq1 and chq2 are from the same quarry, but
two different varieties).
Figure 4.2. Half of the block with holes from the oriented samples. Arrows are oriented the
same way as on the sawn slabs below.
Table 4.1. The table shows the stone types sampled for the major laboratory testing
programme in WP 5, 6 and 7.
code Country of
extraction Rock Type Represented in building No. (Task 1.1) Bowing
Chq1 Calcitic marble DE04 Yes
Calcitic marble
Grq1 Greece Dolomittic marble GR07 No
Itq1 Calcitic marble
Itq2 Calcitic marble
Itq3 Calcitic marble
Calcitic marble
Calcitic limestone-
marble (contact
DE07, DE08, DE09, DK01, DK02, DK07,
DK24, NO03, NO06, NO09, NO10, NO13
Plq1 Calcitic marble PL03 No
Calcitic marble PL01 No
Ophicalcitic marble DE 05, PT05, PT07
Poq2 Calcitic marble DE01, PT02, PT03, PT09 Yes
Poq3 Calcitic marble PT01PT04, PT06, PT08 Yes
Calcitic marble No
Sweden Dolomitic marble SE04, SE06, SE08, SE15, SE16, SE17, SE18,
SE23, SE24, SE26, SE27
Slq1 Limestone (calcitic) SL 02 No
Dolomitic marble SL 01, SL 03, SL 04, SL 05, SL 07, SL 09 Yes
Table 4.2. The table shows the number of samples for screening with the bow test, from
different countries. Each sample consists of several test specimens.
Country No. of samples
Italy 36
Greece 10
Portugal 13
Sweden 8
Norway 7
Greenland 4
Turkey 1
Bulgaria 1
France 1
Total: 86
3.5 WP 5 Research – Finding the mechanism of bowing
3.5.1 Objective
Through field and laboratory investigations, combined with discussions with experts from
industry and research, find the driving force behind and main factors influencing deterioration
due to bowing and/or thermal expansion. Give input to stone companies regarding production
and processing optimising, further developed within WP 8.
The work package is divided into the following four different tasks:
1. Geological framework
2. Production conditions
3. Rock and mineral properties
4. Other properties (moisture gradient etc.)
3.5.2 Task 5.1 Geological Framework (In-Situ Stresses) Objective
To investigate the potential influence of in-situ rock stresses on bowing of marble, three
dimensional rock stress measurements have been carried out in three quarries in the Carrara
area, Italy. Based on the results, laboratory bow tests on specimens cut in various directions
to the major principal stress have been performed on specimens from two of the quarries. Introduction
The original hypothesis for the rock stress measurements was that we would expect high
stresses or high anisotropic stress pattern in areas where strong bowing material is extracted
and low stresses in localities where the marble has not shown problems with bowing. If this
was not achieved, we could conclude that rock stresses were not a major reason for bowing of
marble. In addition, we could expect different responses if the marble is cut in various
directions in relation to the measured principal stress directions, see figure 5.1. The measuring
programme therefore comprised both rock stress measurements and bow tests of specimens
sawn out in different directions according to the main rock stresses. Method
The rock stress measurements have been carried out with a so-called three-dimensional (3D)
overcoring system (figure 5.2). Principally, this is stress relief method, where the rock stresses
are relieved by drilling out a diamond drill core, and where the corresponding rock strains are
recorded by resistance strain gauges. The Young’s modulus and Poisson’s ratio are
determined on the recovered rock core material, and the principal stresses and their directions
are computed using elastic theory.
Figure 5.1. Illustration of potential relation between bowing and direction of rock stresses.
Figure 5.2. Principle of rock stress measurements by 3D overcoring (CSIR overcoring cells).
Figure 5.3. Rock stress measurements in one of the Carrara quarries. Results and conclusions
The in-situ stress measurements at the three quarries clearly indicate that the marble at all
sites is subject to quite high, non-gravitational stresses, and that the major principal stress is
sub-horizontal. Horizontal stresses up to 18 MPa have been measured. The theoretical gravity
stresses are either low or virtually zero (at least at the Gioia and Buca locations).
The stresses are probably a combination of locked-in residual stresses and tectonic stresses.
In the Canaloni quarry, the stresses seems to be relieved the first couple of metres from the
rock face, even if the marble otherwise is seemingly strong and free from fractures. This is
also supported by the fact that no stress indicators like spalling surfaces may be observed.
This may mean that marble blocks taken out probably to a large extent are stress relieved. The
marble has high Young’s modulus (Elastic modulus) and sonic velocity, and has a nearly
perfect linear elastic behaviour, and would under otherwise equal conditions be less prone to
In the Gioia quarry, the stresses are quite high even close to the rock face. This supports
observation of spalling surfaces, and also reports on diamond wire jamming during release of
marble blocks in the quarry. This may mean that relieved blocks still have locked-in stresses
after release from the solid rock. The marble has significantly lower Young’s modulus and
sonic velocity than the marbles from Canaloni and Buca, and also has a more curvilinear
elastic behaviour with hysteresis. This marble would under otherwise equal conditions deform
easier than the two others.
In the Buca quarry, the stresses are quite high close to the rock face, which supports the
observation of spalling surfaces at the measuring site.
The marble has quite high Young’s modulus and sonic velocity, and also shows a linear
elastic behaviour. This marble could be less prone to deform than the Gioia marble.
The results of the actual rock stress measurements show that our original hypothesis is
somewhat simplified. High stresses are measured in all marbles, even in those where no
bowing has been experienced. Nevertheless, for these two quarries, it seems that the stresses
are released during block extraction, in opposite to the quarry where they have experience
with bowing marble. In addition, the rock mechanical properties (Young’s modulus and sonic
velocity) of this marble is much lower than the other two.
As a follow up of the rock stress measurements, bow tests on test specimens cut out parallel
with and normal to the major principal stress direction, and also 45° to these directions were
performed (Figure 5.4 and 5.5). There is a tendency that the highest bowing occur in
specimens cut parallel or 45o to σ1. These findings support a hypothesis that the most
pronounced bowing will occur in directions where the effective shear stresses are highest, due
to the internal friction of the rock.
Average bowing [mm/m], Gioia
0 5 10 20 30 40 50
Number of cycles
Gioia I tq2-par-s igma1-hor
Gioia Itq2-45o-sigma1-hor
Gioia Itq2-135o-sigma1-hor
Gioia itq2-par-sigma1-vert
Gioia itq2-nor-sigma1-vert
Gioia itq2-45o-sigma1-vert
Gioia itq2-135o-sigma1-vert
Figure 5.4. Average bowing values (mm/m) for specimens of Gioia marble cut in various directions
according to the major principal stress .
Average bowing,
0 5 10 20 30 40 50
Number of cycles
Canaloni I tq1-par-sig ma1-hor
Canaloni Itq1-nor-sigma1-hor
Canaloni Itq1-45o-sigma1-hor
Canaloni Itq1-135o-sigma1-hor
Canaloni Itq1-nor-sigma1-vert
Figure 5.5. Average bowing values (mm/m) for specimens of Canaloni marble cut in various
directions according to the major principal stress.
Even though the work has identified a relationship between in situ rock stresses and bowing
magnitude for two marble types, it is difficult to separate this from especially the influence of
rock fabric, i.e. the lattice preferred orientation of the calcite crystals.
In practice it will be very difficult to make use of this information between the rock stress
situation and bowing. Nevertheless, it is important for general block production (yield) to take
the in situ rock stresses present in the rock mass itself into consideration during quarrying. In
order to avoid or reduce the risk of damages to humans, to blocks and to the quarry face itself,
knowledge about the magnitude and direction of in situ rock stresses is important to take into
3.5.3 Task 5.2 Production Conditions Objective and methods
The purpose of Task 5.2 – Production Conditions - has been to find out whether production
and processing practices have any influence on bowing. Establishing practices and conditions
for transformation of raw blocks into finished marble products by site observations and
collecting information from companies has played a major role within this task. Of value has
also been to discuss the path from the selection of a stone material for a specific building
project to the production and processing of the actual stone. Thus, the various parameters,
procedures and practices that can be included in the terms "production and processing" have
been discussed (quarrying and processing methods, influence of cutting directions, panel size
and thickness). In addition, findings from building studies (WP 1 and 2) and other WPs and
Tasks have been brought into this discussion. Laboratory bowing tests on oriented samples, as
well as on specimens with various thicknesses and surface processing (honed, polished, dolly
pointed) have been performed. Results and conclusions
The literature is often vague and insufficient in describing the processes of block extraction
and further processing with respect to bowing and strength loss of marbles. A finding from
the comprehensive literature review performed within the project is that production and
processing parameters are often brought into the discussion when observations are difficult to
explain otherwise.
As an introductory conclusion to the work of Task 5.2 it can be stated that some production
and processing factors influence the bowing and deterioration pattern of marble claddings.
The various factors depend to a large degree on the intrinsic properties of the marble itself.
Thus, one main goal for further use of marble in thin facade claddings should be to seek for
marble varieties with the most favourable intrinsic properties to prevent such changes to
happen. Since various production and processing factors may influence the behaviour of
various marbles to various degree, a second goal is to try to optimise the quality of the final
product, considering e.g. the effect of cutting directions, slab thickness and slab thickness
with respect to surface processing.
Quarrying methods – diamond wire sawing
Several authors have suggested that modern production and processing techniques may
introduce special stresses within the rock material that may enhance the rock bowing
susceptibility. Especially the introduction of diamond wire saws has been suggested as
important in this respect.
There has been a remarkable development in block extraction methods through the last e.g.
50 years. Drilling and blasting has been the traditional method for centuries within the stone
industry. Today, most of the marble quarries in Europe and elsewhere make use of various
sawing techniques, but with drilling with pre-splitting and/or soft blasting as important for
several of the operations in the quarry (final block production etc.).
The age of investigated claddings (concentrated on marble claddings with bowing problems)
ranges from 1 - 270 years. This time interval indicates that several quarrying techniques may
have been used for production of the slabs. From this one could suggest that the extraction
methods can not explain bowing occurring after installation. Another finding from the
building investigations, are that bowing of marble slabs starts "immediately" or at least
shortly after slab installation, and that the deformation reaches a maximum through time.
Since one must assume that the slabs were planar at the time of installation, the bowing
experienced in many marble claddings is thus a result of the material’s response to the
prevailing conditions on the building, and by such very difficult to explain as due to the
extraction method used.
It has been found that various quarrying methods may give various impacts on rocks to be
extracted, but it is very difficult to explain the behaviour in marble claddings as a result of the
quarrying method itself. It is reasonable to assume that drilling and blasting will "cut off"
possible in situ rock stresses in the material, so that it will take a longer time for locked in
stresses to disappear during block extraction in comparison to diamond wire sawing. One
main finding from investigations during diamond wire sawing is that in total a release of rock
stresses in the extracted block is actually happening during this type of block extraction.
Based on this study there is no indication that diamond wire sawing introduces stresses within
the rock material that may enhance rock bowing susceptibility.
Cutting directions
In general one could suggest that the direction of the final building slab is a combination of
(or sometimes a compromise between) the designers request and the producers need for taking
advantage of rock cleavage properties. After the quarrying stage, the blocks to be used for the
building are cut into slabs. The cutting direction may be controlled by the requested visual
appearance, but also by the producers wish to optimise the block yield, selecting a cutting
direction that gives the largest slabs. Thus, in general terms, various cutting directions are
most likely represented in all buildings.
Each material generally shows an optimum cutting direction, in terms of pattern and physical-
mechanical properties. However, this does not mean that it can not be cut in different ways to
obtain special aesthetical features. For marbles with clear, visual foliation this feature is easy
to detect or to gain if requested. Nevertheless, for more homogenous marbles, the cutting
direction may give much of the same visual appearance. However, this does not automatically
mean that the technical properties of slabs cut in various directions are the same.
Both building inspections (figure 5.6) and laboratory bow tests (figure 5.7-5.8) have revealed
that the cutting direction may have an effect on the bowing and expansion potentials if the
fabric of the marble is anisotropic.
Figure 5.6. Interrelation between bowing and cutting direction at part of the Oeconomicum
Building in Göttingen, Germany.
Pecchia marble - Single results. UNIGOE
0 5 10 15 20 25 30 35 40 45 50
# Cycles
Bowing [mm/m
010 20 30 40 50
Temperature cycles [20-80 °C]
Bowing [mm/m]
Figure 5.7. Results of bow tests from UNIGOE. Upper diagram – Peccia (Chq1), lower
diagram – Ruivina (Poq4).
Figure 5.8. Cutting of specimens for bow tests.
Even though the influence of cutting directions on bowing has been studied on marbles with
relative low bowing susceptibilities, there is a tendency that the various cutting directions give
different bowing amplitudes. What has been found from studies within Task 5.3 (see next
chapter), is that both for the Peccia, Ruivina and Rosa Estremoz marbles, which all have a
clear foliation, the grain shape preferred orientation, and thus also the grain boundary
preferred orientation is oriented parallel to the foliation plane and the LPO (texture), i.e. the
preferred orientation of the c-axes of the calcite crystals, is oriented perpendicular to the
foliation plane. Both fabric elements; the grain boundary preferred orientation and the LPO
are the real influencing parameters on the anisotropy of the bowing of marble slabs.
Measurements on building slabs and on slabs cut from fresh quarry blocks have also revealed
high anisotropy of flexural strength for several marbles cut in various directions (figure 5.9).
Cycles (20-80 °C)
Flexural strength [MPa]
Chq1 Poq4 IT
040 0 40 0 40
Figure 5.9. Diagram illustrating the flexural strength as a function of cutting direction for
various marbles (Chq1 – Peccia, Poq4, Ruivina and a Carrara marble). Reference
measurements (0) and specimens after 40 bow test cycles (40). Results from UNIGOE.
The various cutting directions are as follows:
Specimens designed “x”:
Represent specimens cut parallel to the foliation, i.e.
parallel to the xy-plane (blue curves)
Specimens designed “z”:
Represent specimens cut normal to the foliation, i.e.
parallel to the xz-plane (red curves)
Specimens designed “y”:
Represent specimens cut normal to the foliation, i.e.
parallel to the yz-plane (green curves).
Slab dimensions and thickness:
No conclusive answer can be given to the degree and speed of deterioration as a function of
panel sizes, but thickness has been found to influence the bowing and deterioration of marble.
However, there is no "safe thicknesses" of slabs above which bowing will not occur. For a
marble with favourable rock and mineral properties, the behaviour is less dependent on
Surface processing:
Various processing methods give various impacts on stone materials and may change the
microstructure/micro porosity both in and beneath the slab surface. For some techniques this
may influence the capillary properties of the product’s surface and also its strength. It has
been demonstrated that several natural stone varieties, including many marbles and granites
may bow as a response to the processing itself. Both sawing, honing, and bush-hammering
may cause slabs to bow, typically towards the worked face. It is verified that different marbles
respond differently and that there are a relationship between the behaviour during processing
and the mechanical properties of the marble. The measurements indicates that marbles with
high Young’s modulus (stiff) and elastic behaviour will better resist the forces active during
various kinds of processing than less stiff marbles with hysteretic behaviour in the
stress/strain curve during loading/unloading. It should be expected that the thickness of the
slabs are of importance here, the thicker the slabs, the less influenced the material will be by
the processing itself. There also seem to be a tendency that less stiff marbles are more prone
to bow as a result of stresses induced into the material from temperature and humidity
3.5.4 Task 5.3 Rock and mineral properties Introduction
Many studies have tried to explain this phenomenon (e.g. Sir Raiyleigh, 1934; Bain 1940;
Widhalm et al., 1994;Winkler, 1996), and the main environmental factor seems to be
temperature fluctuations in combination with moisture and water that causes the bowing of
the marble claddings. This could be due to the anisotropic thermal expansion of the calcite
crystal (Kessler, 1919) causing an intergranular decohesion of the material (Perrier and
Bouineau, 1997; Malaga 2003).
However, it is important to state that is not all calcite marble claddings that bow, and there are
also cases were dolomite marble claddings displays a weak bowing (Malaga et al., 2004). In
this study we wanted to investigate how the microstructure of the marble influences the
degree of bowing. The microstructure of the marble is dependent of the deformation and
metamorphic history and the recrystallisation process related to these events. A static
recrystallisation causes a grain boundary area reduction, resulting in the presence of even
sized crystals with straight or smoothly curved grain boundaries, forming a granoblastic
texture. When a dynamic recrystallisation occurs, grain boundary migration and recovery of
the material is possible (Passchier and Trouw, 1996). A fabric of dynamic recrystallised
marble is composed of old anhedral grains surrounded by subgrains, forming a seriate
interlobate grain aggregate. This has also been called xenoblastic texture (Royer Carfagni,
Previous studies have shown that there are several microstructural parameters that could
influence the bowing tendencies of marble. Siegesmund et al., 2000 stated that the lattice
preferred orientation as well as the grain fabric control the deterioration of the marble. The
irregularity of the grain boundaries is considered to be another parameter that influences the
deterioration, where an increasing irregularity of the grain boundaries gives a more
sustainable marble (Barsottelli et al, 1998, Royer Carfagni, 1999). The grain size is
considered to be a less important factor for the deterioration of the marble (Zeisig et al.,
Within the scope of the TEAM project, a comprehensive laboratory research programme has
been carried out among the partners. Exposed building samples and samples from quarries
have given a very good spread in types and varieties of carbonatic rocks for the research. The
laboratory analyses have included both standard European tests for general material
characterization and analyses directed towards specific problems. The purposes of the
analyses can be summarized as follows:
1. Characterize and describe investigated rocks
2. Investigate which, how and how much various intrinsic and extrinsic factors influence
3. Define rock properties and test results that indicate durable/vulnerable marbles and
seek threshold values for performance characteristics
4. Study and understand mechanisms for the deterioration
5. Find correlations between test results and real life performance
6. Define which parameters should be tested, i.e. test methods and procedures for
Stone selection for claddings
Product and production control for regular FPC and towards specific building
All rock types from sampled quarries (see chapter 3.4) have been tested and investigated.
Mineralogy has been determined on samples, and sections from all samples have been
quantified by distinct microstructure parameters describing crystal growth, grain boundaries,
grain shapes, grain size, cleavage, twinning, grain size distribution, grain shape factors and
number of neighbours. Lattice preferred orientation of calcite in several marbles has also been
studied. Image analysis has been performed to quantify some of the important microstructure
elements such as grain size distribution and grain shape factors. Adjacent grain analysis has
also performed.
The entire work package contains numerous of marble and limestone types. A summary of all
petrographic analysis is given in table 5.1. Detailed descriptions are given in the technical
report [7].
Table 5.1. Summary of petrographic properties of the main rock types.
Marble Comp Grain boundaries
Shape of grain
growth Twinning D10 D50 D90
Cum80% AG
Chq1 cc Lobate (straight) Interlobe(polyg) Hypidiobl None(weak) Medium 228 508 648 1.70 9
Grq-1 dol Sutured Interlob Xenobl Strong Medium 83 369 591 1.55 10
Itq1 cc Lobate / caries Interlob Hypidiobl Weak Weak 55 119 200 1.70 8
Itq2 cc Straight Polyg Idiobl None Weak(no) 86 173 371 1.50 7
Itq3 cc Sutured Interlob Xenobl Strong Medium 45 127 231 1.70 8
Itq4-1 cc Straight(lobate) Interlob Idiobl None Weak 47 97 174 1.70 8
Itq4-2 cc Straight(lobate) Interlob Idiobl Weak Weak 35 71 123 1.70
Noq-1a cc Lobate / caries Interlob Xenobl Weak Weak 57 159 547 1.50 9
Noq-1b cc Lobate / caries Interlob Xenobl Weak Weak
Poq1x other Lobate(straight) Polyg(interlob Hypidiobl None No 109 387 652 1.70 13
Poq1y other Lobate(straight) Polyg(interlob) Hypidiobl None No 105 416 652 1.60
Poq1z other Lobate / caries Polyg(interlob) Hypidiobl None No(weak) 155 462 647 1.65
Poq2x cc Lobate / caries Polyg Hypidiobl Weak 236 439 629 1.70 8
Poq2y cc Lobate(straight) Polyg Hypidiobl None Weak 275 489 639 1.60
|Poq2z cc Lobate / caries Polyg(interlob) Hypidiobl None No(weak) 265 502 641 1.60
Poq3-1 cc Straight Polyg Hypidiobl Weak No 176 375 555 1.65 7
Poq3-2 cc Sutured Polyg Hypidiobl Medium No 171 401 647 1.65 8
Poq4 cc Lobate (Straight) Interlob (polyg) Hypidiobl None No 104 219 377 1.60 8
Siq2 dol Straight Polyg Idiobl None No 68 126 181 1.60 8
Seq1 dol Lobate / caries Interlob Xenobl None No 133 289 451 1.70 8
M4-1-1 cc Straight(lobate) Polyg Hypidiobl None Weak 97 189 283 1.60 8
M4-1-2 cc Lobate(straight) Polyg Hypidiobl Medium No 82 155 227 1.50 8
M4-2-1 cc Lobate(straight) Polyg Hypidiobl(Idiobl) Weak Weak 101 184 293 1.60 7
M4-2-2 cc Straight Polyg Idiobl(Hypidiobl) None(weak) Weak 87 152 230 1.55 7
M4-3-1 cc Straight(lobate) Interlob Hypidiobl Weak Weak 89 163 249 1.60 8
M4-3-2 cc Straight Polyg Hypidiobl None 81 142 219 1.50 7
The objective of this study was to find a correlation between a microstructural characteristic
that can be quantified in an easy way and the bowing property of marble. This includes also
formulating a reliable and easily applied method that the producers can use as a tool for
quarry planning and the buyers can use as a rapid tool for ensuring a purchase of a suitable
marble for cladding. In order to validate our findings the results obtained on laboratory tested
samples were compared with marble samples taken from buildings.
Figure 5.10. Microscopic images of the samples used for AGA-analysis. Microstructure analysis
After the bow tests samples were cut and vacuum impregnated with epoxy resin containing
fluorescent dye. One thin section, with an area of approximately 1200 mm2, was made from
each sample. The microscopic images in figure 2 show that there are clear differences in the
microstructure for the investigated samples. The microstructure was quantified as follows:
The 2-dimensional grain size distribution was measured and the 3-dimensional grain size
distribution was calculated. In addition, an adjacent grain analysis was applied, see section below.
62 Grain size distribution
Traverses were randomly drawn on microscopic images. The maximum ferret diameter was
then measured on each mineral cutting a traverse. The measurements of the ferret maxima
only express the 2- dimensional grain size distributions of the samples. The NT Build 486
method (Sandström, 1995) was used in order to calculate the 3 dimensional grain-size
distributions. For natural sand this method reproduces the results obtained through sieving
(Lindqvist & Sandström 1999). Adjacent grain analysis - AGA
A calcite crystal belongs to the hexagonal crystal system. In an ideal even grained
granoblastic texture, all calcite crystals share grain boundaries with six grains. These are in
the following referred to as adjacent grains AG. An increasing complexity of the grain
boundary and a more heterogeneous grain size distribution, changes the relationship and
increase the number of adjacent grains. In a calcite marble with a heterogeneous grain size
distribution, the largest crystals have the highest number of adjacent grains whereas the
smallest crystals can have less than six adjacent grains. The same features occur in a calcite
marble with complex grain boundaries of the crystals. Both of these microstructures increase
the ratio between the total grain perimeter and the square root of the number of analysed
grains in comparison with a granoblastic texture. This relation was observed by Bain (1941)
in an investigation on grain-border measurements on several types of marbles.
The marbles used in this study range from almost perfectly granoblastic (Ve1) to seriate
interlobate (Bi5 and Bi6). In order to quantify the increasing complexity of the microstructure
were the number of adjacent grains counted around the measured mean sized grains in the
rock. The reason for using only medium sized grains is that if the adjacent grains are counted
around the smallest or the largest mineral grains would the results not be representative. The
selection of analysed grains was determined from the grain size measurements. The AGA was
performed on images taken with a polarisation microscope using a digital camera and
enlarged on the computer screen to easily count the number of adjacent grains. The AG was
counted around one hundred grains on each sample and the results are shown in figure 5.10. Results
The results from the grain-size analyses show that the samples form two groups. One group
that have between 70-80 % of counted grains less than 63 μm according to the 3D-grain size
calculations. The two other samples have no counted grains less than 63 μm (figure 5.12). The
same groups are observed in the ferret max plot in figure 5.11.
The results from the AGA (table 5.2) show median values from 6 to 9 for the number of
adjacent grains, and the distribution of AG are plotted in figures 5.11 and 5.12. Samples Bi5
and Bi6 that showed the lowest bowing tendencies have 9 AG, whereas sample Ve1, showing
a bowing of 1,4 mm/m have 6 AG. The samples with 7 to 8 adjacent grains (sample Bi1 and
Bi2) have a more irregular grain shape than Ve1, showing a granoblastic texture and they
have also a more heterogeneous grain size distribution (Fig 5.10).
Feret max (um)
Figure 5.11. 2D-grain size distribution of the investigated samples.
Numberbased 3D-distribution
Grain size (um)
Figure 5.12. 3D-grain size distribution of the investigated samples.
Table 5.2. Adjacent grain analysis for the investigated samples
Sample No of AG (median values)
Bi 1 7
Bi 2 8
Bi 5 9
Bi 6 9
Ve 1 6
The results from the quantitative analyses of the microstructure showed that there is a good
correlation between the type of texture, grain-size distribution and the number of AG.
Additionally, the microstructure seems to have an important influence of the magnitude of
bowing. Those samples that have 9 AG show a microstructure referred to as seriate interlobate
texture (figure 5.10). In low magnification under the microscope, the grain boundaries look
rather complex and very irregular. However, in high magnification it is possible to see that
these boundaries only show a slight suturation, which is almost the same feature as grain
boundaries in a granoblastic texture (figure 5.13). The "irregular grain boundaries" are instead
represented by several small, 5-20 μm euhedral to subhedral grains. These observations
indicate that the shape of the grain boundaries is not the crucial parameter. What seems more
important for a durable marble is the amount of fine-grained matrix around the larger mineral
grains. This is in agreement with Zeisig et al., (2002) who showed that marbles with irregular
grain boundaries could show the same residual strain as marbles with a granoblastic texture.
Figure 5.13. Microphotograph of sample Ve 1 (left) and Bi 5 (right). The images show that
the grain boundaries of Bi 5 only, are rather sutured and the "irregular grain boundaries"
observed in lower magnification and shown in figure 5.10 is instead represented by several
small grains. In sample Ve 1 is no fine-grained matrix observed. The image surface
corresponds to 333x250 microns.
To evaluate the relevance of the results from the laboratory tests, the AGA technique was
applied on samples from four different buildings, which are included in the TEAM project.
One building is situated in Denmark and three in Sweden (Fig 8). One of the buildings shows
no bowing whereas the three other show bowing up to several centimetres per meter. Figure 9
shows the typical microstructure from these marble claddings. It also shows that the three
buildings with bowing panels all have granoblastic textured marble whereas the non-bowing
façade cladding have a similar microstructure as sample Bi5 and Bi6. The results from the
AGA were comparable with those obtained on the laboratory samples (table 5.3).
Table 5.3. Adjacent grain analyses and bowing from the four buildings.
(Measurements from Yates, 2004, except Sydsv, Alnaes et al., 2004).
Building No of AG Bowing (mm/m)
Nyk (Sweden) 6 20-30/strong
Sydsv (Sweden) 6 30-40/very
Mch (Sweden) 9 1-2/none
RK (Denmark) 7 10-20/medium
The sample from Denmark differs from all other investigated samples. The grain size
distribution of this marble is very heterogeneous with grains of a few microns up to several
mm, (actually cm size in some places) but these grains are not mixed together, they rather
form clusters of fine and coarse-grained areas. A determination of the grain-size distribution
alone is therefore not a sound basis for assessing the bowing properties of a calcitic marble.
Figure 5.14. The four investigated buildings. Nyk show a bowing of 20-30 mm/m; Sydsv; 30-
40 mm/m; Mch 1-2 mm/m; RK, 10-20 mm/m.
Figure 5.15. The microstructure of the marble claddings from the four buildings. Three of the
images show an almost granoblastic texture similar to Ve 1, whereas sample Mch has a
microstructure similar to Bi5 and Bi6. The image of building RK also shows that this marble
has a very heterogeneous grain size distribution. Conclusions
The microstructural properties have been quantified using several techniques, including
adjacent grain analyses, AGA. This technique is a fast and easy method to make a numerical
description of the microstructural transition from granoblastic to seriate interlobate marble.
The results show that the marbles with granoblastic texture all have six adjacent grains (the
median value). They also show that with increasing amount of fine-grained matrix and, to
some extent a more heterogenous grain size distribution, the number of adjacent grains
increased. Additionally, the results also show that there is a good correlation between the
number of adjacent grains and the degree of bowing. The samples with six adjacent grains
showed the greatest degree of bowing.
The present investigation only deals with pure calcite marbles. There are known cases where
dolomitic façade claddings display weak bowing. Dolomite does not have the same
anisotropic characteristics concerning e.g thermal expansion as the calcite crystal, and should
therefore be treated as an individual group.
3.5.5 Task 5.4 Other Properties (moisture gradient etc.) Objective
The purpose of this section of the report is to demonstrate the research that underpins the two
tests methods developed and evaluated in WP 6. In particular, it shows the processes by which
the parameters and their values were established.
Before Introduction and background
The main purpose of this task was to find out which external environmental and use factors
influence bowing and/or expansion of marble and limestone; how they influence the bowing
and/or expansion; and by how much. Based on the findings from WP 1 and 2, preliminary
laboratory analyses (direct bowing tests and test methods for irreversible thermal and hydric
properties) were performed in order to determine critical levels of moisture and moisture
gradients; sample conditioning properties; and temperature intervals. The findings from these
screening tests were used to develop and refine the "bow test" and the "expansion test" that
were evaluated in the inter-laboratory comparison in WP 6, task 6.1.
In addition, the implications of anchoring systems (WP 2.6) on bowing/expansion were
considered but at the time that the tests were being developed there was no evidence for any
direct relationship and so this was not included in the test methods evaluated in WP 6.
There have been many studies of the bowing and deterioration of marble and these are
reviewed in the State-of-the-Art Report produced as part of WP 1. Deterioration of marble
panels involves several parameters and properties. Bowing is the most obvious phenomenon,
but bowing is often followed by a volume change, i.e. the marble expands. However, for
obvious reasons the most serious deterioration feature is the loss of strength which may
progress so far to total loss of cohesion (decohesion) between the grains. The above three
features may cause cracking of panels, spalling in connection with anchor points and in the
worst cases ultimately failure of the panel.
Even though no conclusive explanation for the bowing of marble had been found prior to the
TEAM project, two parameters were agreed on: temperature variations and moisture, in
combination. The project has developed from this point, and other key factors like the internal
microstructure of marble has now been recognised (Alnæs et al 2004). Grelk et al. (2004)
pointed out that, in theory, laboratory testing for bowing, expansion and flexural strength is a
very good method to establish the durability and like future behaviour of a given marble.
Successive in-situ measurements and inspections of field exposure sites (Malaga et al. 2004)
and of building facades (Yates et al. 2004) have contributed to the knowledge on the bowing
mechanisms in relation to environmental conditions and construction specific parameters.
These findings have been used to underpin the development of two laboratory tests – one for
bowing of marble and the other for the permanent expansion of marble.
Table 5.4. Selected stone types from quarries- Material data.
Break out load at dowel holes
code Country of
extraction Rock type Commercial name Porosity (Hg)
Porosity (%)
(weight %)
Different orientation versus the foliation
Chq1 Calcitic marble Peccia Virginio Normal 0,33 - - - - 10,4 1,51 1,55 1,53
Peccia Colombo - - - - - 15,0 1,78
Grq1 Greece Dolomitic marble Snow White of Thassos 0,30 0,67 2824 0,22 0,72 13,2 2,61
Itq1 Calcitic marble Venato 0,07 0,30 2708 0,10 0,59 21,1 2,06
Itq2 Calcitic marble Bianco Ordinario 0,63 0,57 2703 0,20 0,50 7,7 1,26
Itq3 Calcitic marble Bianco Ordinario 0,23 0,38 2704 0,13 0,64 22,9 2,29
Italy Calcitic marble (Breccia) Arabescato 0,42 0,60 2704 0,21 1,50 15,5 1,86
Norway Calcitic-silicic limestone
(contact methamorphic)
Porsgrunn 0,21 0,46 2708 0,18 0,60 13,0 1,91 2,42 2,33
Plq1 Calcitic marble Biala Mariana - - - - - 17,5
Calcitic marble Slawn-iowice (“white”) - - - - - -
Poq1 OphiCalcitic marble Verde Viana 0,35 - - - - - 1,58 1,57 1,69
Poq2 Calcitic marble Trigaches 0,18 - - - - - 1,78 1,86 1,81
Poq3 Calcitic marble Rosa Estremoz 0,12 0,39 2703 0,15 0,54 9,9-22,5 2,09 2,16 2,57
Portugal Calcitic marble Ruivina 0,18 0,54 2696 0,16 0,79 11,4 1,75 1,95 1,99
Seq1 Sweden Dolomitic marble Ekeberg 0,15 0,31 2859 0,15 0,36 21,0 2,47
Slq1 Limestone (Calcitic) Hotavlje Grey - - - - - -
Dolomitic marble Sivec - - - - - -
Figure 5.15. Correlation between strength parameters and bowing of marble slabs (from
Royer Carfagni 1999).
0 5 10 15 20 25 30 35
Number of cycles
Weight increase
Figure 5.16. Expansion and weight increase of marble prisms exposed to a number of heating
cycles (20-80 oC) in wet conditions. The weight increase is measured in "wet" condition and
corresponds to an increase of porosity in the marble. Observations from visual surveys of buildings
One of the first tasks of the project was to collect data examples of bowed marble cladding
from Europe. This data collection has continued throughout the project and has been
expanded to include examples from the USA. A total of 194 buildings have been identified
and described at various levels. 26 of these buildings have been selected for more detailed
investigation, of which 6 of them have been selected and investigated both with detailed field
study and some measurements of bowing. All of the 26 buildings are considered as suitable
for dismounting of facade panels for laboratory testing purposes. Involved partners have tried
to get samples dismounted from the buildings for laboratory testing purposes. The selection of
buildings for WP 2 (6 buildings) and WP 3 (3 buildings) was also chosen among these 26
buildings. To date examples have been recorded from around 50 different locations in Europe
but it is likely that many more examples remain un-noticed.
There is, for natural reasons, a concentration of buildings affected by bowing reported from
countries facing on the Baltic Sea – particularly northern Germany, Denmark, Finland and
Sweden, but there are also examples from the rest of Europe including Slovenia, Switzerland,
Austria, France, UK, Belgium, Italy and Poland. This is merely the reflection of the partner’s
initial knowledge, and ease of access to sites and information. The most well known example
in Europe is the Finlandia Hall in Helsinki, Finland. There is no evidence that any particular
climate is typical of the conditions that can result in long term bowing and expansion. Cases
of bowing have been reported from the most different climates, from Libya in the south to
northern Sweden/Norway in the north but a temperature range and a source of moisture are
common to all locations.
A wide range of buildings were visited and the condition recorded along with details of the
type of stone used, the panel dimensions, the fixing system and the local environment. The
main findings from this stage of the work were:
Facade compass direction and height over ground
Bowing is observed to all facade directions as well as on all heights over ground. On same
facade section there is a tendency that bowing amplitudes are more significant higher up on
the facade. There is also a tendency that facades facing south and west have higher bowing
amplitudes then is the case for facades facing north and east.
Colour of marble
Among reported projects different colour marble types with bowing problems are included
such as light types from Italy, Switzerland, Russia, Macedonia and Sweden, green type from
Portugal, dark grey types from Portugal and Norway. It shall be stressed that same types of
marble on other reported projects does not show any bowing. Thus there is no clear indication
as to the effect of colour range of the marble related to deformation by bowing or not bowing.
Panel face dimensions and thickness
Very large marble panels (> 2m2) have been recorded with perfectly plane and unaffected
surfaces while on other buildings small (< 0,1 m2) have deformed, deteriorated and fall from
the facade. No correlation has been observed between bowing tendencies and stone panel
thickness. It might however be noted that marble panels on Nyköping City Hall (Sweden)
which are 30 mm thick have deformed in concave direction while adjacent smaller panels
which are 60 mm thick on the same facade elevation have deformed in convex direction.
Convex or concave bowing
It has not been possible to relate the options of convex or concave to any other influencing
factor. Panels on one side or one location of the same building might deform in concave
direction while on an other side or location the bowing is convex, for example at Lünen
Hospital (Germany), Nyköping City Hall .
It has also been noted that the original marble facade panels on Finlandia Hall in Helsinki
(Finland) were mainly concave with extremely high amplitudes after 30 years of exposure.
However, the panels installed some 3-4 years ago of same white Italian marble type (however
not from same quarry) have started to deform in a convex direction.
Another striking example is Torsviks Torg in Lidingö (Sweden) – here three tower buildings
which are covered on all sides with marble panels (600 x 900 x 30mm) show convex and
concave bowing on alternate rows of panels.
Anchoring system
Most of the inspected facades have 30 or 40 mm thick marble panels on various types of
anchors and with a ventilated air cavity 20 – 40 mm wide between the stone panel back face
and layers of insulation. Some facades are, however, installed on mortar beds with or without
restraint anchors – this means that there is no ventilated cavity. Some projects have part of the
installation done on freely standing railing systems allowing both the front and rear panel
faces to be "exposed" to the surroundings. To date no link has been observed between the type
of anchoring system and the bowing of marble panels with the exception panels which are
bedded on mortar which do not seem to be susceptible to bowing.
Building measurements
Some preliminary measurements were made at a number of these buildings using a modified
version of the "bow meter" that had been developed in the earlier Nordtest project (see NT
Build 500). The "bow meter" is basically a 1200 mm straight edge with a digital dial gauge
that allows the distance from the edge to the panel surface to be measured very accurately.
Detailed site investigations, measurements, sampling etc, have, so far, been performed at 8
buildings (seven are recorded in table 5.5, the eighth is the University Library at Göttingen,
Germany). In addition, panels were removed from 10 buildings for further investigations in
the laboratory, seven of these have been completed to date. Petrographic thin sections have
been produced and described from all these samples and they have been subjected to a
laboratory bowing test and the flexural strength determined. A summary of the results of the
site measurements and laboratory bowing is given in table 5.3.
Table 5.5. Summary of the results from site measurements and laboratory tests on seven
buildings in Europe.
Building name Country Marble type Bowing
Nyköping City
Sweden Italian
20-30mm / strong
Lünen Hospital Germany Portuguese
30-50mm / very
Bank Building,
Denmark Norwegian 10-20mm / medium
Office Building,
Denmark Norwegian 10–20mm / medium
Malmö City Hall Sweden Italian
1-2mm / none
Hotel Terazza Sweden Swedish
5-15mm / medium
Lyngby City Hall Denmark Greenland
0-2mm / weak
Probably the most important physical change observed is the loss of flexural strength. The
results of changes in strength recorded during the literature review and the testing of samples
from buildings are summarised in table 5.5. In addition to the bowing and loss of strength
there are also examples of spalling of stone around the fixing points (often associated with the
movement of the panel against the fixing) and erosion of the surface as the marble becomes
less durable.
The survey confirmed that although buildings are found in a wide range of climates they are
linked by having significant temperature ranges and wetting and drying cycles. This helped to
confirm the basic parameters for the laboratory tests – temperature and moisture cycling. The
survey also showed that panels could respond to these parameters by expanding and that in
cases where the faces expanded at different rates that this resulted in bowing. It was also clear
that the samples used in the laboratory tests need to be of suitable dimensions for flexural
strength testing since this would provide a key part of the data required to establish likely
service life.
Table 5.6. Summary of recorded losses of flexural strength.
Age Loss of flexural
(Year) (from literature)
Trigaches E./ 14 ~30%
Escamado 28 ~75%
Bank building in
Copenhagen (Danish
National Bank)
Denmark Porsgrunn
23 ~ 45% Leksø
~ 35%
Bardiglio ~ 10%
Office building in
Sweden 20 ~ 10%
Office building in Paris France ~ 10 -15
Office building in Paris France 11 ~ 50%
Office building in
France 5 ~ 50%
Office building in
Denmark 60 ~ 45% Leksø
Office building in
Denmark 41 ~ 75%
Collegiata di
Italy ~ 100 ~ 90% Garzonio
Office building Switzerland 3 ~ 40%
(2001 b)
Hospital, Lünen Germany Stocksiefen
Office building in
Sweden 31 ~ 75%
Amoco USA 15 ~ 40% Hook (1994)
Finlandia Hall Finland 21 ~ 85%
Building Country Marble
type Reference
Figure 5.17. Distribution of buildings where bowing or distortion of the marble cladding has
been recorded by the TEAM partners. Development of Laboratory Screening Tests
The development work was designed to build on the test method NT Build 499, developed
prior to the TEAM project (see Nordtest NT Build 499). This uses a standard test specimen of
400 mm length, typically 100 mm width and a thickness, similar to the panel’s thickness (or
30 mm in standard tests). The specimen is placed in an insulated container, where it is placed
on a tray, filled with a layer of filter cloth or sand. The tray is filled with distilled or
demineralised water up to approx. 10 mm below the upper surface of the test specimen.
The specimens are exposed to a number of cycles. Each cycle begins with an exposure to
infrared heating from above, which lead to an increase of the surface temperature from the
ambient room temperature to 80 °C over a period of 1-3 hours. The surface temperature is
maintained at 80 °C for 2-3 hours, after which and the specimen allowed to cool to ambient
room temperature, (20 °C) until at least 24 hours has passed from the start of the exposure
cycle. Temperature measurements on panels on buildings by Perrier & Boineau (1997) and as
part of this project have revealed that the daily magnitude of the temperature cycles can be up
to 60 °C and that most of this temperature increase happens in within 3 hours. The bowing of
the specimen is measured after a number of cycles.
The test method has been applied on a large number of different stone types during the TEAM
project and has yielded very important results. In addition, the maximum temperature, the rate
of heating, and the amount of moisture present were all varied during the development work
in the TEAM project. The bowing will usually grow unlimited in an environment where there
are temperature and moisture cycles. The same tests have been carried out on test specimens
from the same sample, but without any water in the tray, corresponding to a dry exposure.
Figure 5.13 shows that the bowing after a few cycles reaches a stable level in a dry
environment, after which the bowing does not increase. This illustrates that bowing only
becomes critical in environments, where moisture is also present.
The bowing potential differs significantly between different stone types, and the thickness of
the panel will also have an effect on the development of bowing.
Figure 5.18. Bowing on Norwegian marble versus cycles; wet and dry exposure.
A: Dry exposure until 56 cycles, then wet.
B: As A, but with wetting of upper surface.
C: Wet exposure for the first 56 cycles, then dried.
D: Wet exposure for the first 56 cycles, then constant temperature
in dry environment.
E: Wet exposure.
F: Wet exposure, but sample is turned after 56 cycles.
Bowing and loss of strength have been reported by Yates et al (2004) to correlate in samples
from building claddings. Flexural strengths have therefore also been measured on a number of
bowing laboratory specimens. A correlation of loss of flexural strength and bowing has been
observed on the laboratory specimens.
One of the first to report irreversible thermal expansion of different marble types was Kessler
(1919), but many researchers have later shown, that repeated heating cycles lead to permanent
expansion of marble. The TEAM-project has therefore developed a test method for permanent
expansion of marble, exposed to temperature cycles.
This test method uses a standard test specimen of 30 x 30 x 300 mm. The temperature of the
water in the tank follows the same variation as the surface temperature in the bowing test with
a tolerance of +5 °C. The length of each specimen is measured 2 hours after 80+5 °C in the
water has been reached. The temperature is then decreased to the ambient temperature and the
length is measured again at least 2 hours after 20+5 °C in the water has been reached.
The expansion can be mapped as a function of the number of exposure cycles as shown in
Figure 5.19, where it can be seen that the expansion seems to grow continually with the
number of exposure cycles. The similar exposure can also be carried out in a dry environment
and will lead to an expansion, which after a few cycles reach a permanent level. This
difference between the wet and the dry exposure corresponds to the differences observed in
the bowing testing. The expansion testing can also be carried out with every second cycle dry
and every second one wet, simulating an environment, where moisture is only available in
some periods. Figure 5.19 shows that this exposure (wet/dry) will lead to approx. the same
expansion as a constantly wet environment.
The permanent expansion may lead to large forces in the panels and joints and may lead to
failure of the panels or the fixings and so it is important that the magnitude of the likely
expansion can be established in the laboratory.
Figure 5.19. Permanent expansion versus cycles; for wet and dry exposure of Norwegian
marble. In-situ site monitoring observations
Selection of monitoring parameters and sites
Based on the wide range of geographical locations, materials and buildings identified and
recorded in WP 1, three sites were selected for continuous monitoring of deformation/
movements for panels and monitoring of environmental conditions in order to determine
under which conditions bowing is occurring and how the changes vary with time. This
necessitated the development of equipment for long term monitoring, installation of
equipment at three locations and processing of the results. The data is being used to calibrate
model of the risks of failure associated with the changes in the properties of the panels over
long periods of time in order that these risks can be better understood and managed.
The selection of parameters to be measured, selection of equipment and selection of suitable
buildings were all given careful consideration before being agreed. It was agreed that ideally
at each location monitoring would include:
surface temperature on the external surface of the stone
time-of-wetness/condensation on external surface of the stone
strain in two directions on the external surface of the stone
surface temperature on the internal surface of the stone
air temperature in the gap behind the panel
relative humidity in the gap behind the panel
time-of-wetness/condensation on internal surface of the stone
strain in two directions on the internal surface of the stone
shade air temperature
shade relative humidity
There were problems with finding ideal sites and so there was a need to "compromise". The
final list of sites was:
Danish National Bank, Denmark
University Library Göttingen, Germany
Nyköping City Hall, Sweden
Installation of equipment
The first two sets of equipment were installed in Copenhagen at the beginning of September
2001, two sets of equipment were installed in Göttingen in November 2001, and three sets at
Nyköping in October 2002. The "sophistication" of the equipment has progressed as the team
has gained further experience of the sites and have overcome the initial problems relating, for
example, to the fixing of the strain gauges.
Danish National Bank
Amongst the most interesting results from the Danish National Bank in Copenhagen is the
fact that there is very little difference between the temperatures on the front and back of the
panels – only a few minutes "lag". On the eastern side the front can be 4 ºC warmer when the
sun first shines on this face. The full temperature rise can take 6 hours but in more extreme
cases the rate is around 0,3 °C per minute and this information has been used in the
development of the bowing test.
Göttingen University Library
The equipment has been in place since November 2001. The initial aim of the logging at this
site was to evaluate a new type of temperature and humidity logger and to evaluate the first
strain gauges for their thermal stability and the magnitude of movement. Most of the
equipment seems to be working well but there are concerns about the strain gauges as the data
shows sudden changes in strain. One set of covers have been added but it does not seem to
have solved the problem. Despite some problems with the downloading of data interesting
results have been found with diurnal cycles clearly visible against a background of longer
term changes with reliable stable data being obtained over short periods – for example 7- 8
However, over longer periods the "drift" and background strain readings still seem to show
sudden changes.
Nyköping City Hall
The equipment was installed on the City Hall at Nyköping in October 2002 – this was timed
to coincide with the setting up of the field exposure site. The aims at this site are to evaluate
the re-designed strain gauges, in particular to look at the magnitude of movement, expansion
vs bowing and the effects of surface treatments.
The results show that the panel which was removed from the building in November 2002 and
replaced on new fixings shows a fairly rapid bowing immediately after the panel was replaced
– possibly as it moved back against the fixings- followed by a long period with very little
One panel was of new Carrara marble and this has bowed in the opposite direction to "old"
panel and the magnitude of the bowing is also much greater.
The third panel at Nyköping was new Carrara marble that had been treated with a
hydrophobic coating. The treatment was intended to reduce the bowing. Initially the rate
seemed to be reduced but this became less clear after a longer exposure period.
The in-situ monitoring provided important information for the development of the tests –
particularly data on the likely temperature ranges (both diurnally and annually) and the
temperature gradient through the thickness of the panel. Summary and conclusions
The purpose of this Task was to consider and quantify a wide range of data from earlier
research, observations on buildings, in-situ monitoring, and laboratory tests in order to verify
that the two laboratory tests evaluated in WP 6 provide a relevant and realistic prediction of
what would happen to the panels under the field conditions in a structure. The results of the
testing of the bowing potential and the expansion potential must therefore be correlated to the
observations on structures.
The work has confirmed the importance of external factors – particularly
temperature variations in combination with humidity are the external factors
required for bowing to occur.