Combadura y dilatación de paneles de piedra natural : ensayo y evaluación de màrmol y caliza

Natural stone has been used as a building material for centuries. In the past, load bearing members were made of entirely of stone, but in the last 50 years new processing techniques have made the production and use of thin facade cladding a profitable venture. Unfortunately however, marble facades on buildings in Europe and elsewhere have undergone severe deterioration. The EC-financed TEAM project (2000-2005) studied the bowing observed on marble facades in both cold and warm climates. TEAM’s main objectives were to understand and explain the expansion, bowing, and strength loss mechanisms governing the decay of marble- and limestone-clad facades, and to draft new European standards to prevent the use of marble and limestone poorly suited to outdoor cladding. A survey of some 200 buildings afforded a clear picture of the geographical, geological and climatic scope of the problem. Detailed case studies of six buildings resulted in a facade assessment methodology that included a monitoring system and risk assessment. Both laboratory and field research was conducted on almost 100 different types of stone from different countries and in place in different climates. The outcome was the determination of the decay mechanisms and critical factors. Two test methods and respective precision statements, one for bowing and the other for irreversible thermal expansion in high humidity conditions, were prepared for submission to CEN TC 246.


SUMMARY
Natural stone has been used as a building material for centuries.In the past, load bearing members were made of entirely of stone, but in the last 50 years new processing techniques have made the production and use of thin facade cladding a profitable venture.Unfortunately however, marble facades on buildings in Europe and elsewhere have undergone severe deterioration.The ECfinanced TEAM project (2000)(2001)(2002)(2003)(2004)(2005) studied the bowing observed on marble facades in both cold and warm climates.TEAM's main objectives were to understand and explain the expansion, bowing, and strength loss mechanisms governing the decay of marble-and limestone-clad facades, and to draft new European standards to prevent the use of marble and limestone poorly suited to outdoor cladding.A survey of some 200 buildings afforded a clear picture of the geographical, geological and climatic scope of the problem.

INTRODUCTION
Natural stone has been used on facades for centuries.Originally, the dimension stone used in structural members was cut very thick, ensuring highly consistent strength and durability.Scientific research on the properties of marble began in the late nineteenth century.In the years that followed, the thickness of the natural stone used on facades decreased from over 1000 mm (for structural members) to 20-50 mm (for cladding) thanks to the new cutting technologies and equipment developed by the quarrying industry.The use of thin marble and limestone slabs for facade cladding has grown substantially over the last few decades.Although the vast majority of marble-clad facades perform satisfactorily, durability problems began to occur with the advent of thin cladding materials.And indeed, the number of reports of facade failures has increased dramatically in recent years.
Emblematic structures such as the Amoco building in Chicago, the Finlandia City Hall in Helsinki and La Grande Arch have all been affected by severe flaws in their thin marble-clad facades.Flaws include expansion, bowing and strength loss, and in the most severe cases complete detachment from the anchoring system.At present, most of the reports of facade failure are located in Europe and North America.Flaws have also been recorded in other parts of the world, however, such as Lebanon, India and Cuba.Despite the research conducted in Europe, the primary solution consists in replacing the panels, at a cost of approximately €500-1000 /m².A fairly recent example can be found in Helsinki's City Hall, all of whose panels were replaced in 1998 at a cost of nearly €4 M.This provides an idea of the economic dimension of the problem.Although most of the reports on the unduly short durability of thin marble or limestone panels involve Italian Carrara marble, the most widely used marble in the world, bowing has also been reported in connection with American, Norwegian, Portuguese and other types of marble (Figure 1).Moreover, the reports on Carrara marble performance are inconsistent, for in some cases it apparently performs satisfactorily.Panels may bow either convexly or concavely to the facade.
The TEAM project consortium included 16 partners from 9 EU countries who represented stone producers and trade associations, testing laboratories, standardization ciantes de la piedra, laboratorios de ensayos, organismos de normalización y certificación, consultores, propietarios y responsables del mantenimiento de edificios y fabricantes de sistemas de reparación.Se centró en el problema de la dilatación y la combadura del mármol y de la dilatación de la caliza en revestimientos exteriores (1).

OBJETIVOS Y CONTENIDOS DEL PROYECTO
Los principales objetivos del proyecto eran: • Conocer y justificar los mecanismos de la dilatación y de la pérdida de resistencia.
and certification bodies, consultants, building owners, maintenance service firms and repair system manufacturers.The project focused on the problem of expansion and bowing in marble and expansion in limestone cladding for outdoor use (1).
Its results are the outcome of a review of over 300 articles, a survey of around 200 buildings, personal knowledge and experience and comprehensive laboratory and field work.It produced findings that have proved useful to stakeholders such as the stone industry, standardization, testing and research organisations and building owners, as well as consultants and architects.
The existing hypotheses (2-5) were compiled and evaluated.A number of authors had suggested that the anisotropic thermal expansion of calcite crystals may underlie intergranular decohesion in marble (6-9).Siegesmund et al. maintained that both the preferred lattice orientation and the grain fabric control marble deterioration (9).Grain boundary irregularity, in turn, was regarded to be able to prevent or decrease bowing (10)(11)(12).Frost action was commonly assumed to impact bowing substantially.Grain size had been reported to have only a minor effect on bowing (13).Many of these hypotheses were tested under the TEAM project.

PROJECT OBJECTIVES AND CONTENT
The main objectives of the project were: • To understand and explain the mechanisms governing expansion and strength loss.• Encontrar tipos de mármol con tendencia a la combadura.
• To prevent the use of unsuitable marble and limestone by introducing drafts for European standards on the subject.
• To determine whether surface coating and/or impregnation could prevent or diminish degradation.
Among the most common hypotheses/assertions verified were: • Carrara is a poor quality marble.
• Fine-grained marble is suitable for use as cladding.
• Moisture and temperature variations are crucial to bowing.
• Anisotropic thermal expansion of calcite and dolostone causes granular decohesion.
• The release of stress locked into rock plays an important role.
These hypotheses were analyzed in greater detail on six buildings, two in Sweden, two in Germany, one in Italy and one in Denmark.The parameters studied in the project included: • A portable device called a "bow-meter" was developed specifically for the project to take high precision measurements in the field (Figure 3).Bowing was measured four times a year and then correlated with the strain measurements.The data gathered contributed to the definition of the climatic variables for the laboratory bow and moisture expansion tests and to a better understanding of the diurnal variations in panel temperature.By fil- • Temperatura superficial en la superficie interna de la piedra • Temperatura del aire en el hueco entre el panel y el muro • Humedad relativa dentro del hueco entre el panel y el muro • Temperatura del aire a la sombra • Humedad relativa a la sombra Para el proyecto se desarrolló un dispositivo portátil llamado "combímetro" (Figura 3) para medir con gran precisión la combadura sobre el terreno.Se midió la combadura cuatro veces al año y luego se estableció una correlación con las mediciones de deformación.Este seguimiento generó datos para la definición de las variables climáticas en el ensayo de combadura del laboratorio y el ensayo de dilatación en húmedo, así como un mayor conocimiento   de las variaciones de temperatura diarias de los paneles.La filtración de estos cambios a pequeña escala permitió establecer una buena correlación entre las mediciones de la deformación residual producida a largo plazo y las mediciones manuales de la magnitud de la combadura realizadas con el combímetro.Una conclusión muy importante fue que la combadura varía hasta 2 mm/m en el mismo día.Las mediciones con combímetro realizadas de forma aislada pueden ofrecer valores poco claros, por lo que deben hacerse de forma repetida y en condiciones climáticas parecidas.
El desplazamiento diario de los extensómetros podía registrarse de forma inequívoca con el equipo.También estaba claro que el desplazamiento diario probablemente "enmascarara" algún desplazamiento residual producido más a largo plazo en los paneles a raíz de la dilatación o combadura iniciales del mármol.La Figura 4 muestra el panel 1 (panel original), que fue desmontado de la fachada y vuelto a montar con nuevos anclajes y equipado con dispositivos de medición.La pauta de desplazamiento parece indicar una combadura bastante rápida justo después de volver a montar el panel (posiblemente porque éste se estaba ajustando a los nuevos anclajes), seguido por un largo periodo de poco desplazamiento.La Figura 5 muestra los resultados del panel 3 de mármol de Carrara Gioia.Obsérvese que el panel se está arqueando en la dirección contraria a la del panel 1 y que la magnitud de la combadura también es muy superior.La Figura 6 muestra los resultados del panel 2 de mármol de Carrara Gioia, que se había tratado con un recubrimiento hidrófobo (pro-tering these small-scale changes a very good correlation could be established between the long-term residual strain readings and the manual bow-meter measurements.One very relevant finding was that bowing varied by up to 2 mm/m during a single day.As the information from single bow-meter measurements can be confusing, repeated readings should be taken, always under similar climatic conditions.

DATA ANALYSIS AND INTERPRETATION
Selected TEAM project results are discussed in this paper.
More detailed studies and all the results, including the methods developed, can be found in the final report of the project ( 1).
An evaluation of the data gathered at Copenhagen and Göttingen led to the design of more stable and reliable equipment for the Nyköping measurements.Consequently, the following discussion is based on the Nyköping findings only.The 35-year-old Bianco Carrara marble cladding on this facade exhibits significant bowing.Three panels were monitored in detail: one original panel and two new Carrara Gioia marble replacement panels.The new Carrara Gioia marble was selected for the experimental studies due to its strong bowing properties detected in TEAM screening tests.
The daily movements measured by the strain gauges were large enough not only to be recorded by these instruments, but very likely to 'mask' any longer term residual movement in the panels ensuing from early expansion and/or bowing of the marble.Panel 1 (shown in Figure 4), the original panel, was removed from the facade, fitted with the measurement devices and set back in place with new anchors.The readings seem to indicate fairly rapid bowing immediately after the panel was remounted, possibly as it adjusted to the new anchors, followed by a long period with very little movement.The results for panel 3, made with Carrara Gioia marble, are given in Figure 5.Note that this panel bowed much more than and in the opposite direction from Panel 1. Panel 2, likewise made of Carrara Gioia marble, was treated with a water-repellent (anti-graffiti) coating (Figure 6).The aim was to assess the capacity of the coating to reduce ducto antigraffitti).Con el tratamiento se pretendía evaluar la posibilidad de que redujera la combadura.Los resultados mostraron una reducción de la velocidad de combadura horizontal y vertical de aproximadamente el 50% debida a la impregnación, en comparación con el panel 3.
bowing.Both horizontal and vertical bowing in this panel was found to be close to 50% lower than in Panel 3.
El grosor de los paneles influye en la velocidad de combadura y deterioro.Los análisis de paneles de 20 mm, 30 mm y 40 mm de un mármol con el mismo tipo de combadura demostró que los paneles de 20 mm fueron los más sensibles a la combadura y a la desagregación granular, mientras que los paneles de 40 mm de grosor eran más resistentes a la combadura.
Para algunos tipos de mármol se observó una relación entre la estructura y textura de la roca (orientación predominante del esqueleto) y la resistencia a la combadura y la dilatación.La dirección de corte de un panel puede por tanto influir en la combadura y la dilatación de los mármoles con una marcada anisotropía.Se analizaron cerca de 100 muestras con intención de averiguar la posible relación entre la orientación predominante del esqueleto y la combadura.La influencia sobre la magnitud de la combadura fue bastante baja en el caso de los tipos de mármol con una orientación bien definida.Es importante destacar que la diferencia era inferior a la precisión del método de ensayo.Sin embargo, al ser sistemática no podía pasarse por alto.Debe por tanto señalarse que los

DISCUSSION
The similarity found between the strain gauge monitoring data and the bow-meter measurements indicated that the "real time" strain gauge measurements could be used to supplement and corroborate the manual bow-meter measurements.
Panel thickness was observed to influence the bowing and deterioration rates.Analyses of 20-, 30-and 40-mm panels of a type of marble prone to bowing showed that the 20-mm pieces were the most sensitive to bowing and sugaring.Sugaring is a popular term for granular decohesion.Bowing was retarded in the 40-mm panels.
The rock fabric (preferred lattice orientation) was found to be related to bowing/expansion (strength) in some types of marble.The direction in which a panel is cut may therefore affect bowing and expansion in highly anisotropic marbles.Nearly 100 samples were studied to determine the potential relationship between preferred lattice orientation and bowing.The impact of lattice orientation on bowing was very small for clearly oriented marbles.Although the differences were in fact smaller than could be measured with any precision using this test method, they appeared too systematically to be ignored.It should be stressed, therefore, that marbles poorly suited for outdoor Figura 6. Panel 2 de Nyköping: combadura horizontal y vertical (V) de un panel nuevo de mármol de Carrara tratado con un recubrimiento antigraffitti.tipos de mármol inadecuados para el revestimiento de exteriores no deberían recomendarse con independencia de la orientación elegida a la hora de extraer y elaborar las piezas.
La posibilidad de comparar los resultados del laboratorio con las observaciones realizadas en los edificios fue muy valiosa, aunque muchos resultados y relaciones obtenidos en el laboratorio no pudieron probarse sobre el terreno (un mismo mármol se arqueaba en una misma fachada de forma cóncava y convexa pero en condiciones de laboratorio lo hacía en sólo una de las direcciones).
Sawing, honing and bush-hammering may cause slabs to bow, typically due to expansion in the surface involved.Different marbles responded differently, and a relationship was found between behaviour during processing and the mechanical properties of the marble.The most vulnerable appeared to be low stiffness marbles whose stress/strain curve is characterized by hysteresis during unloading.Such marbles also tend to be more prone to bowing as a result of stress induced in the material by external factors such as temperature and humidity gradients.
The other tests conducted included water absorption, capillary suction, flexural strength, yield strength, frost resistance, ultrasonic pulse velocity, long-term bending, heat capacity, heat conductivity and specific heat.One of the first and most relevant findings was that marble and limestone samples necessarily had to be dried at temperatures lower than normally specified.Most standards specify that stone samples should be conditioned at 70 ºC.The TEAM results showed that some types of marbles underwent a significant reduction in strength even at temperatures lower than 70 °C.Drying at 40 ºC for one week is recommended to minimize damage to the specimens prior to testing.This criterion was followed throughout the project.
The opportunity to compare laboratory results to developments observed on buildings proved to be extremely useful, although many of the laboratory findings could not be confirmed in the field: one and the same marble observed to bow both convexly and concavely on a given facade, for instance, bowed in one direction only under laboratory conditions.
While granular decohesion is commonly believed to be the sole cause of expansion and bowing, this decay mechanism was not observed in all the types of marble studied.It was found more often in calcitic than dolomitic marbles, although many pure calcitic marbles showed no signs of granular decohesion.Differential expansion in different crystal lattice orientations is a property that marble shares with many other rock types, granodiorite among them, and although it does not in itself explain bowing, it is a factor to be considered.
Se confirmó que los factores intrínsecos más importantes son la complejidad de los bordes de los granos y la granulometría de los granos de mineral de la roca (Figura 7), que están relacionados con las distintas fuerzas de unión entre los granos de mineral debido a la complejidad de la disposición de sus bordes, conjuntamente con la estructura cristalina.Este tipo de estructura permite una mayor adherencia entre los granos debido al contacto más fuerte que supone desde el punto de vista mecánico y una superficie de contacto mayor en comparación con los granos "ideales" de forma hexagonal regular.El tipo de grano de mineral o cristal con formas complejas y defectos de estructura se adapta con mayor facilidad a los cambios mediante la redistribución de átomos individuales que los granos ideales y sin defectos de forma hexagonal regular.Esto se debe a que dichos cristales tienen una energía interior mayor que los de forma ideal.Sus defectos son el resultado de tensiones internas acumuladas durante una recristalización dinámica que pueden liberarse sin afectar a los bordes de los granos.La acción de un ciclo de temperatura y humedad intenso sobre una fachada puede desencadenar estas reacciones.Los cristales de forma ideal no tienen la misma capacidad para redistribuir los átomos dentro de su esqueleto y, por tanto, están limitados a movimientos a lo largo de los bordes de los granos, provocando la desagregación granular.Por otra parte, la propagación de fisuras y la dilatación y coalescencia de los huecos puede producirse con mayor facilidad en una estructura con bordes de granos rectos.
No se conoce con toda claridad la influencia del agua, pero se sabe que en los ciclos de humectación y secado se acumulan tensiones.En los mármoles con capacidad potencial de combadura los bordes entre los granos son más abiertos que en los materiales que no se arquean, por lo que entra más agua haciendo que actúen a micro- The crucial parameter was shown to be the complexity of grain boundaries and the mineral grain size distribution in the rock (Figure 7).The differences in bonding strengths between mineral grains were related to the complexity of the contours of grain boundaries, in conjunction with the structure of the crystals themselves.Such structures provide for stronger inter-grain bonding thanks to the greater mechanical strength of the interface and because the area involved is greater than in "ideal" hexagonal grains.Mineral grains/crystals with complex shapes and lattice defects are more prone to change via redistribution of individual atoms than flawless, ideally shaped hexagonal grains because the former have higher levels of inner energy than the latter.Such defects are the result of internal stress built up during dynamic re-crystallisation that can be released without affecting grain boundaries.
Release may be triggered by exposure to the high temperatures and moisture cycles prevailing on facades.Since ideally shaped crystals cannot rearrange atoms within their lattice as readily as irregularly shaped grains, the response of straight-sided crystals to such stress is essentially confined to movement along grain boundaries, with the concomitant granular decohesion.In addition, smooth grain boundaries facilitate crack propagation as well as void expansion and coalescence.
How water affects the process is not completely clear, but stress is known to build up during wet-dry cycles.
Marbles with bowing potential have more open grain boundaries than non-bowing types of stone.Such a structure facilitates water uptake and increases microscale capillary forces during both wetting and drying, ren-Figura 7. (a) microtextura no arqueable de mármol con bordes de granos complejos y gran variedad de tamaños de granos; (b) mármol con tendencia a la combadura debido a su estructura relativamente simple con granos aproximadamente del mismo tamaño y bordes de granos rectos.
Figure 7. Micro-texture in (a) non-bowing marble with complex grain and widely varying grain sizes; and (b) marble prone to bowing, with its simpler structure, smooth grain boundaries and more uniform grain size.

(a) (b)
dering the marble unsuitable for outdoor use.Moreover, the energy required to initiate and propagate cracks is lower in wet than dry rocks because it takes more additional energy to replace a solid-solid contact with a solidair contact than to replace it with a solid-water contact.Consequently, strength is lower in wet than in dry stones.
In addition, an open microstructure facilitates water uptake and the concomitant frost damage in the winter, further accelerating decay.Open structures are also more vulnerable to weathering as a result of the acid in polluted rain.

BOW TEST AND MOISTURE EXPANSION TEST
That a moisture gradient is needed for bowing to occur was an assumption adopted early on in the project.Such warping had often been observed in flooring, where marble tiles bowed soon after being laid in the grout.The bowing observed in tile floors was not permanent, however.By contrast, the defects observed on buildings was largely permanent as a result of a combination of high temperatures and a moisture gradient.A laboratory test was therefore devised to simulate these conditions.This called for creating the same conditions as observed in facades.The information needed about the maximum temperatures reached on stone facades (at least in Europe) was gathered from the detailed building inspections and long-term monitoring conducted, as well as the review of the literature.A temperature of 75 ºC was measured on a dark, vertical limestone panel.The normal temperature on a white marble facade on a sunny summer day is around 55 ºC.These studies also afforded information on "ramp time", i.e., the time needed for panels to reach the maximum temperature and to subsequently cool down.One important consideration addressed when designing the method was that the temperature cycle would have to be repeatable in every test, and equal "climatic" conditions ensured for all test specimens regardless of type, colour and other characteristics of the marble.This could only be achieved by using a black reference as described in ISO 4892-1 (Plastics -Methods of exposure to laboratory light sources).Note that this procedure controls the climate, but not the stone temperature.
escala fuerzas capilares mayores en el caso de los mármoles inadecuados, tanto durante la fase húmeda como durante la seca.Asimismo, la energía necesaria para que se formen y propaguen las fisuras es inferior en una roca húmeda que en una seca.Esto se debe a que el aumento de energía es mayor cuando se sustituye un contacto de sólido con sólido por un contacto de sólido con aire, que cuando se sustituye un contacto de sólido con agua.
Conviene destacar que mediante este procedimiento es el clima el que se controla y no la temperatura de la piedra.
The second trial developed was an expansion test that correlates more closely to performance observed in the field than any of the standard methods presently available.Building designers generally prefer very small joints between stone panels.Such "expansion" joints should ideally accommodate movements between the individual panels and thereby prevent breakage.In the past, these joints were engineered around the "thermal expansion coefficient" in dry conditions.Building facades are not always dry, however, and expansion in the presence of moisture is usually significantly higher than in dry conditions.Consequently, the test was conducted on watersaturated samples.
The two test methods were tested in a comparative trial to the requirements laid down in ISO 5725-1994 Basic Methods for the Determination of Repeatability and Reproducibility of a Standard Measurement Method.Repeatability and reproducibility were calculated accordingly.

Bowing test
Since tests conditions were not identical in all laboratories, the results varied more than expected.Both the stones and the bowing method proved to be very sensitive to differences in temperature, and not all the laboratories were in a position to install the black reference during the time the project lasted.Slight differences in equipment setup were also reflected in variations in the test results.Nonetheless, all the laboratories were able to successfully discriminate between bowing and non-bowing marble with this test.The test apparatus is shown in Figure 8.
No obstante, las fachadas de los edificios no siempre están secas y la dilatación en condiciones de humedad es por lo general considerablemente mayor que en condiciones secas.En consecuencia, en el presente trabajo, el ensayo se ha basado en muestras saturadas de agua.
Ambos métodos de ensayo se utilizaron en una prueba comparativa de conformidad con los requisitos recogidos en la norma ISO 5725:1994 Basic Methods for the Determination of Repeatability and Reproducibility of a Standard Measurement Method.El cálculo de la repetibilidad y reproducibilidad se realizó de acuerdo con dicha norma.
One obvious difference was observed between the in situ bowing direction and the laboratory results.While in practice bowing direction is practically arbitrary, in the bowing test the specimens always curved toward the heat (i.e.upwards).These findings, which are difficult to explain, have also been reported by other authors (14,15).

Expansion
A series of practical difficulties had to be surmounted to conduct the expansion test.For instance, despite the many trials conducted with different types of glue, during exposure to hot water, instruments tended to work loose from the stone surface at some of the measuring points.
The initial number of test specimens should therefore be large and care should be taken to ensure that the glue hardens.This is a problem common to many expansion tests, concrete and mortar tests among them.Like the bow test, the expansion test can be further refined, but in its present state it has proven to be able to determine whether a given type of marble or limestone is apt for cladding.Furthermore, it provides valuable information for dimensioning expansion joints.Another relevant observation was that certain types of marble and limestone tended to expand continuously, while in others expansion was constrained.Heat-induced expansion is illustrated in Figure 9.
Figura 9. Ensayo de laboratorio de dilatación por humedad y calor a fin de obtener datos fiables para el dimensionamiento de las juntas de dilatación.El recipiente de agua con temperatura controlada se emplea habitualmente en los ensayos de cemento.Ambos métodos de ensayo se completaron mediante la determinación de la resistencia a flexión tras los ciclos de temperatura, con la que se obtiene una indicación clara de la potencial disminución de la resistencia con el paso del tiempo, así como valiosos datos para predecir la vida útil.Sin embargo, debe tenerse en cuenta que en caso de necesitarse un cálculo más preciso de la vida útil, será necesario conocer la temperatura y las variaciones de temperatura de la superficie de la piedra en la ubicación real del edificio.Si se ha seleccionado un mármol propenso a la combadura, también será necesario determinar la resistencia a las heladas de la piedra antes y después de los ciclos de temperatura.

CONCLUSIONES
A partir de un examen exhaustivo de la literatura y el estudio de un conjunto de edificios se pudo definir con gran certeza el alcance geográfico y geológico del problema.La combadura es un fenómeno mundial que no se limita a un determinado tipo de mármol o un clima en concreto, y que puede producirse sin que intervenga la acción del hielo.El análisis de una serie de casos prácticos permitió explorar varias hipótesis sobre el deterioro y proporcionó información importante sobre las variables que deben utilizarse en los métodos de ensayo desarrollados como parte del proyecto, especialmente en lo referente a la gama de oscilación de las temperaturas.
Los resultados del seguimiento a largo plazo revelaron la importancia de realizar las mediciones repetidas veces para conseguir e interpretar los datos de manera relevante y fiable.Las mediciones realizadas una sola vez sirven de poco debido a las grandes variaciones diarias y estacionales.
El muestreo de los materiales de ensayo es fundamental para todo proyecto.En este sentido, son imprescindibles instrucciones detalladas sobre la toma y el marcado de las muestras.El CEN TC 246 sobre piedra natural ha sido informado de los resultados del presente trabajo y las recomendaciones correspondientes.
El ensayo en laboratorio de casi 100 tipos distintos de mármol ha permitido rebatir muchas de las hipótesis existentes.El principal factor extrínseco que influye en la combadura y la dilatación es la temperatura elevada en presencia de un gradiente de humedad, que crea las tensio- The above two tests were supplemented by determining flexural strength after the temperature cycles.This procedure provided a clear indication of the potential strength decrease over time, as well as data relevant to predicting service life.More precise estimates of service life would be possible, but would depend on in situ information on temperature and temperature fluctuations on the stone surface.Where bowing marble is selected, its frost resistance may also have to be determined before and after the temperature cycles.

CONCLUSIONS
A comprehensive review of the literature, together with a survey of affected buildings, provided a very good picture of the geographical and geological scope of the problem.Bowing, which occurs around the world, is not confined to any one type of marble or climate, and frost action is not requisite to its appearance.A series of detailed case studies not only afforded the opportunity to test several hypotheses on deterioration but furnished relevant information on the variables to be used in the tests developed for the project, in particular as regards appropriate temperatures.
Further information about deterioration and the rate at which it takes place was gathered by sampling and testing building panels.This, together with the inspection methodology, constituted a sound basis for formulating a model to predict the remaining service life, including an analysis of the associated risk.
Long-term monitoring highlighted the importance of repeated measurements for the relevant and reliable interpretation of the data.Single measurements were shown to be of little use due to wide diurnal and seasonal fluctuations.
Sampling test materials is of vital importance to any project.Sampling, in turn, depends on detailed instructions both on the sampling process itself and sample marking.The findings and recommendations ensuing from this study have been reported to CEN TC 246 Natural Stone.
Many "old" hypotheses were refuted by the results of laboratory tests on nearly 100 different types of marble.The external factor with the greatest impact was found to be high temperature in the presence of a moisture gradient.These conditions were seen to generate the exter-nes externas a las que los distintos mármoles responden de distintas maneras.Los factores intrínsecos más importantes son la complejidad de los bordes de los granos y la granulometría de los minerales que constituyen la roca, que, junto con la estructura cristalina, proporcionan distintas fuerzas de adherencia entre los granos.La estructura de grano irregular, algo que tienen en común todos los mármoles considerados adecuados para el revestimiento de exteriores, es el resultado de la combinación del metamorfismo que convirtió la caliza en mármol y una recristalización dinámica.Las uniones más débiles provocan la desagregación granular del mármol y pérdidas de resistencia significativas.
nal stress that different marbles respond to in different ways.The crucial intrinsic parameter, in turn, was the complexity of grain boundaries and the mineral grain size distribution in the rock.Indeed, the differences in bonding strengths between mineral grains were due to the complexity of grain boundary contours, along with the structure of the crystals themselves.Such irregular grain structures, a feature that all marbles regarded to be suitable for outdoor cladding have in common, is a result of the combination of two factors: the metamorphism whereby limestone was converted to marble and dynamic recrystallization.Weaker bonds lead to granular decohesion or sugaring in the marble as well as significant strength loss.
The results of the bowing and expansion tests proved that every marble is unique and responds uniquely to climatic stress in accordance with its own degradation curve.Consequently, the acceleration factor determined with the laboratory bow test differs from one type of marble to another.The tests developed provide for a reliable assessment of marble suitability, or otherwise, for outdoor cladding.The bow test can, moreover, be adapted to predict the remaining service life of a given marble on a specific building.
Field trials showed that marble decay can be inhibited or reduced by coating the surface with a water-repellent substance.
Many of the misconceptions reflected in previous projects and reports may be due to the complexity of the problem.Testing only one or very few types of marble without validating the results against actual case studies has often resulted in erroneous or unsubstantiated conclusions.
The TEAM project results have enhanced the understanding of marble and limestone deterioration considerably.Marble can be used for outdoor cladding, but should be thoroughly tested before installation.The project findings should be applied to ensure the selection of suitable stone, thereby contributing to the safe use of marble and limestone as cladding materials.

Figure 1 .
Figure 1.Expansion-mediated concave and convex bowing of marble panels.
Cladding design and anchoring systems • Climatic conditions, facade orientation and building physics • Above grade elevation • Panel cracking and breaking • Open or sealed joints; joint width • Fabric/orientation of natural stone foliation • Convex or concave bowing • Surface finish and treatment Monitoring equipment was developed and installed on three of the six buildings: Danish National Bank in Copenhagen, Denmark, University Library in Göttingen, Germany and City Hall in Nyköping, Sweden (Figure 2).The following parameters were monitored: • Surface temperature on the outer side of the stone • Time-of-wetness/condensation on the inner and outer sides of the stone • Strain in two directions on the inner and outer sides of the stone • Surface temperature on the inner side of the stone • Air temperature in the gap behind the panel • Relative humidity in the gap behind the panel • Air temperature in the shade • Relative humidity in the shade

Figure 4 .
Figure 4. Panel 1 at Nyköping: horizontal (H) and vertical (V) bowing in an original Carrara marble panel after removal and remounting.

Figure 6 .
Figure 6.Panel 2 at Nyköping: horizontal (H) and vertical (V) bowing in a Carrara marble replacement panel treated with an anti-graffiti coating.

Figure 8 .
Figure 8. Laboratory bow test apparatus for combining a wet underside with cyclic heating from above.

Figure 9 .
Figure 9. Laboratory test for moisture-and heat-induced expansion, to obtain reliable data for dimensioning expansion joints.The thermally controlled water bath is standard cement testing equipment.