Monument Future

Sandstone types and specimens

Seven types of sandstone were excavated in the past directly in Prague or in mostly close Bohemian quarries.

They are denoted by the names of the quarries and include Božanov (arkose sandstone), Žehrovice (arkose), Droba (wacke sandstone, unknown quarry), Hořice (quartz sandstone), Libná (quartz sandstone with glauconite), Praha (quartz sandstone), Petřín (quartz sandstone).

84

Figure 1: Divison of the prismatic samples taken from rail stones that had to be replaced.

Prismatic samples in dimensions of 50 mm × 50 mm × approx. 200 mm were cut into two test specimens – a weathered part with the degraded surface layer and a part of the unweathered material, Figure 1. The deteriorated stone exhibited a significant variation of its characteristics along the depth profile. Therefore, the stone specimens were prepared first in the form of cubes for non-destructive US tests, Figure 2.

Then the cubes were cut into thin plates according to the methodology recommended by Drdácký & Slížková (2008). Thin plates enable to design a sequence of tests that provide first data from nondestructive tests, typically volumetric change due to hydric and temperature variations, and then from destructive tests of mechanical characteristics, Figure 3.

The procedure above was applied to the non-weathered specimens as well as on the weathered stone with both the uncleaned and cleaned deteriorated surfaces. For the cleaning, a sandblasting approach has been adopted.

Figure 2: Ultrasonic testing of material characteristics in 5 mm equidistant profiles.

Figure 3: Scheme of the cutting plan for characterization according to the depth profile of the sandstone specimens.

Sandstone consolidation

For the pilot consolidation tests, two ethyl silicate-based agents have been selected, namely non-diluted Funcosil^® Steinfestiger 100 and Funcosil^® Steinfestiger 300. They were applied in amounts of 1l per 1 m² after one-week conditioning at 20 °C/60 % RH. The agents were applied by syringe extended with the cigarette filter on stone surfaces vertically arranged in order to imitate the expected treatment situation on the rail walls, Figure 4.

85

Figure 4: Treated stone specimens with marked points for US velocity measurements and clearly visible depth of penetration.

After the treatment, the specimens were conditioned for one month at 20 °C/60 % RH before starting the testing.

Material testing

During experimental work, the following characteristics were tested: ultrasonic velocity in transmission, micro-drilling resistance, water uptake, porosity, hydric dilation and thermal dilation, bending strength, modulus of elasticity and frost resistance.

Test results

Porosity and mechanical characteristics represent the most interesting data at the consolidation tests. Changes of porosity, as well as in mechanical characteristics, substantially influence the behavior and life cycle of treated historic materials. Naturally, the surface stone deterioration creates very non-homogeneous profiles along the depth in different distances from the surface. Stone material then responses in various ways to consolidation interventions (Sasse & Snethlage 1996). Figure 5 illustrates changes in US velocities in the tested sandstone after consolidation with the Funcosil 300 Steinfestiger.

Figure 5: US velocity changes after consolidation by Funcosil 300 – non-weathered stone (upper) and weathered stone (lower).

The highest increase of ultrasonic velocity was observed in samples from Praha sandstone (light blue lines), which has the highest porosity as well as the largest mean pore size. Other stones appeal improvement in USV only in case of weathered surface treated by the higher concentration of consolidant.

86

Figure 6: Three-point bending tests of thin sandstone plates.

Figure 7: Changes in the bending strength of the unweathered sandstone after consolidation.

Similarly, the bending strength tests on thin plates (Figure 6), cut from the unweathered stone samples, exhibited the highest impact of consolidation on the high porosity Praha sandstone, Figure 7.

In most cases, the consolidation by Funcosil 100 caused a higher increase in strength than by Funcosil 300.

Figure 5 above clearly shows significant differences in the material characteristics of the weathered and deteriorated sandstone in the depth profiles. Due to crust formation on some stones, the surface and near-surface layers may have elevated mechanical properties – strength and the modulus of elasticity usually together with a decreased mean pore size. On the other hand, disintegrated sandstone types exhibit lower mechanical properties and higher mean pore size characteristics. As an example of both types, let us present Figure 8 showing a variation of the bending strength in the depth profile and Figure 9 comparing mean pore size variations.

Figure 8: Examples of the bending strength variations according to the dept profile of the weathered and “virgin” sandstone samples.

Figure 9: Comparison of mean pore size before and after consolidation.

87

Figure 10: Consolidation impact on the stone porosity.

Figure 11: Hydric dilatation after Funcosil 300 treatment on the first two plates under the treated surface – αH in μm/m.

It is seen in Figure 8 that the bending strength of the inner layers of the weathered stone is higher than that of the virgin material. Here must be taken into account that the weathered layers might have some consolidation history, which is not exactly known but could increase the strength of the original material in the near-surface layers.

The weathering with subsurface deposits, as well as the consolidation, decreases the volume of pores especially by filling the small pores which reflect in an increase of the mean pore size value. The value of porosity changes can be studied on Figure 10.

From the other test results, the hydric dilation changes are interesting. Figure 11 shows a series of results of hydric dilation measurements on the surface and the first subsurface layers of the weathered stones. At the same time, the effect of sandblasting cleaning has been investigated.

In Figure 11 the dark blue denotes the weathered uncleaned material, the light blue the material which was sandblasted.

It is apparent that the cleaning of the stone surface significantly reduced hydric dilatation up to 5 mm depth. Probably some effect of packing during blasting may be the reason.

In depths from 5–10 mm (second plate), the hydric dilatation is more affected by a consolidation agent.

Conclusion

The tests were required by restorers before planning a rather massive conservation campaign on the Charles bridge in Prague – one of the most important stones Gothic structure. The results achieved helped to make an appropriate choice of consolidation agent, to decide about necessity and type of surface cleaning, to be prepared for a selection of an appropriate stone in cases of replacement needs and to assess intervention impacts. It enhanced the overall design of restoration interventions.

Acknowledgements

The paper is based on the results of research supported by the institutional project RVO 68378297. The authors acknowledge experimental support of E. Čechová, A. Zeman, J. Valach and professional advice of J. Novotný.

88References

Drdácký, M. F., Slížková, Z. Performance of glauconitic sandstone treated with ethylsilicate consolidation agents, In Proc. of the 11th Int. Congr. on Stone, Vol. 2. Toruń; 2008. pp.1205–1212.

Sasse, H. R., Snethlage, R. Evaluation of stone consolidation treatments. Science and Technology for Cultural Heritage. 1996;5(1):85–92.

Table 1: Annex. Sorption characteristics of the tested sandstones.

89

THERMAL BEHAVIOR OF BUILDING SANDSTONE: LABORATORY HEATING EXPERIMENTS VS. REAL FIRE EXPOSURE

Nadine Freudenberg¹, Thomas Frühwirt¹, Klaus-Jürgen Kohl², Monika Kutz², Heiner Siedel³, Jörn Wichert³

IN: SIEGESMUND, S. & MIDDENDORF, B. (EDS.): MONUMENT FUTURE: DECAY AND CONSERVATION OF STONE.

– PROCEEDINGS OF THE 14TH INTERNATIONAL CONGRESS ON THE DETERIORATION AND CONSERVATION OF STONE –

VOLUME I AND VOLUME II. MITTELDEUTSCHER VERLAG 2020.

1 Technische Universität Bergakademie Freiberg, Institute of Geotechnics, Gustav-Zeuner-Str. 1, 09599 Freiberg, Germany

2 Institut für Brandund Katastrophenschutz Heyrothsberge, Research Division, Biederitzer Str. 5, 39175 Biederitz, Germany

3 Technische Universität Dresden, Institute of Geotechnical Engineering, 01062 Dresden, Germany, Heiner.Siedel@tu-dresden.de

Abstract

Heat-induced short-term decay of dimension stone on buildings and monuments caused by fire is a well-known phenomenon. Most of the scientific studies about thermal behavior and thermal changes of building stones are carried out in laboratory ovens by stepwise heating of stone samples to different stages of temperature. However, real conditions of fire attack on stone elements of buildings might differ considerably from the relatively slow, even heating of small samples in ovens. Therefore, more realistic fire scenarios were designed to test the behavior of sandstone specimens such as cylinders and balusters (height 58 cm and max. diameter 19 cm). The samples comprise the Cotta and Posta type of the Cretaceous Elbe sandstone. They were exposed to a real scale fire test, based on the standard ISO 9705 (room corner test). The specimens were mounted in a fire container at a height of 170 cm above the fire source, a wood crib in accordance to DIN EN 3–7. The standard defines a known theoretical heat release rate, producing a maximum air temperature of approx. 900 °C for about 15 minutes. The temperature in the container as well as on the surface and within the stone specimens was monitored by thermocouples during the tests. The measured surface temperatures vary between 350 and 600 °C, whereas the temperatures at some 4.5–9.5 cm below surface vary only between 200 and 350 °C, depending on the shape of the samples. After the fire tests, different crack patterns were observed. In contrast, smaller specimens heated in a laboratory oven did not reveal any macroscopic cracks, although they were exposed to the same or even markedly higher temperatures (1,000 °C in the sample core). However, both treatments are needed for a better understandig of fire damages on stone buildings since the material behavior of sandstone on grain size scale (fabric and mineralogy) triggers macroscopic crack patterns such as fragmentation and scaling.

Introduction

Due to firestorms caused by heavy bombardments during the Second World War, lots of buildings and objects made of sandstone were massively damaged. These damages became probably more severe by extinguishing fire by water, leading to another short-term temperature change. The typical damage observed is fragmentation, thus reducing the load-bearing capacity of architectural elements, as shown by the examples of the Church of Our Lady (Frauenkirche) and the altar in the Church of the 90Three Kings (Dreikönigskirche) in Dresden, Germany (Fig. 1).

Figure 1: Fire damages of historical buildings in Dresden a) Church of Our Lady (Frauenkirche) 1971 (Wikimedia Commons, CC BY-SA 3.0, Lencse Zoltán) b/c) altar in the Church of the Three Kings (Dreikönigskirche).

Numerous studies (e. g. Chakrabarti et al. 1996, Hajpál & Török 2004, Gómez-Heras et al. 2006, Hager 2014, or Lintao et al. 2017) deal with methods to record material changes of different sandstones caused by high temperatures. However, most of these studies investigate the thermal behavior of small samples with laboratory heating regimes in high temperature ovens. In contrast, there are only few studies dealing with small scale real fire scenarios of sandstones, e. g. Koser & Althaus (1999), Ehling & Köhler (2000), Pohle & Jäger (2003), McCabe et al. (2007), or Smith & Pells (2008). Obviously, the damage patterns of heat-treated laboratory samples and fire-affected objects and buildings (cf. Fig. 1) are different. This study on Elbe sandstones compares the behavior of oven-heated with flame-treated samples, the latter corresponding to a more realistic fire scenario.

Materials and testing procedures

The investigated material comprises sandstone of Cotta and Posta type which are the two main varieties of the Upper Cretaceous Elbe sandstone, occurring south of Dresden (Saxony, Germany). The Cotta type is a grey to yellowish-brownish sandstone with clay-bearing, organic and ferritic flakes parallel to bedding. It is a fine-grained and siliceous quartz arenite (> 90 % quartz). In addition, K-feldspar, kaolinite and few illite, glauconite, and rare organic components occur. The color of the Posta type varies between light grey and yellowish-brownish. It is a fineto medium-grained, occasionally coarse-grained, porous and siliceous quartz arenite (quartz nearly 100 %) without organic matter and with very small amounts of kaolinite (Grunert 2007, Grunert & Szilagyi 2010).

For the laboratory heating experiments cylindrical specimens of both sandstone types with different dimensions of 50 × 25 mm and 50 × 100 mm (Fig. 2a) were used. They were orientated normal and parallel to bedding. The dimensions of the specimens for the real-scale fire exposure tests were significantly larger with heights of 58 cm and an approx. diameter of 19 cm. To imitate real shapes of architectural elements such as pillars, balusters and cylinders were carved from Cotta and Posta type sandstone blocks (Fig. 2b/c).

The small sandstone specimens (cf. Fig. 2a) were treated in a laboratory oven (Nabertherm LT24/12) at the Institute of Geotechnics, Technische Universität Bergakademie Freiberg (TU BAF) at 6 different temperature levels (400, 500, 600, 700, 800, 1,000 °C) with a heating rate of 10 K/min and a cooling rate of 1 K/min after a holding time of 6 hours at each target temperature level.

The cylinders and balusters (Fig. 2b/c) were marked for drilling boreholes to mount thermocouples (Fig. 3a/b) which monitored the temperatures on the stone surfaces and within the stones during fire exposure over time. In the cylinder samples, 5 boreholes with a diameter of 8 mm were drilled 91to a depth of 9.5 cm. In accordance to the specific shape of the balusters, 7 boreholes with a diameter of 8 mm were drilled to depths between 4.5 and 9.5 cm. Flowable mortar was used to fix the thermocouples in the boreholes and to guarantee undisturbed heat transfer.

Figure 2: Investigated specimens of Posta and Cotta type Elbe sandstone a) small cylinders (50 × 100 mm) b) cylinders (58 × 19 cm) c) balusters (58 cm in length and max. diameter of 19 cm).

Figure 3: a) Scheme of drill holes on a baluster and a cylinder specimen b) cylinders with mounted thermocouples.

For the real scale room fire tests a fire container (height: 2.40 m, width: 2.35 m, depth: 4.13 m) was used based on the standard ISO 9705 (room corner test at the Institute of Fire Protection and Disaster Control (IBK) in Heyrothsberge (Fig. 4). In the fire container, the cylinder and baluster sandstone specimens were placed at a height of 1.7 m above the fire source (Fig. 5a), achieving a direct flame treatment. The fire source consisted of a wood crib according with DIN EN 3-7 which provided a known theoretical heat release rate with a maximum temperature of approx. 900 °C for about 15 minutes. N-heptane acted as a fire accelerant which was ignited in a pan below the wood crib.

The temperatures in the container were monitored by thermocouples over time. An infrared and a video camera (Fig. 5b) recorded the heat distribution and the fire behavior in the container which could be followed in real-time on a monitor in the nearby laboratory (Fig. 5c/d).

Figure 4: Sketch of the fire container (top view) with sample and thermocouple positions; red: wood crib in accordance to the DIN 3-7 standard as fire source.

92

Figure 5: a/b) Views inside the fire container b) positions of video and infrared camera c) record of video monitoring d) record of infrared monitoring.

Results and Discussion

The macroscopic results for the stone samples after heating are shown in Fig. 6a/b. There are significant differences between heating in the laboratory oven and in the fire container. In Fig. 6a the small Posta and Cotta type sandstone specimens with axis parallel (PS_P & CS_P) and normal (PS_N & CS_N) to bedding are displayed. They are arranged according to their temperature treatment levels (from left to right: 25, 400, 500, 600, 700, 800 and 1,000 °C). The specimens appear more reddish with higher temperatures. These color changes are related to mineral transformations, namely of iron-bearing minerals (cf. Hajpál & Török 2004).

Within the brownish to yellowish Elbe sandstones, mainly limonite changes to the red hematite at elevated temperatures (Fig. 6a). Slight color changes to red can be detected for all sandstone types already at 400 °C (cf. Gómez-Heras et al. 2009). In Cotta sandstone also glauconite transformations might contribute to discoloration.

The cylinders treated in the oven did not reveal any macroscopic cracks. The treated and untreated specimens were stored in plastic bags after cooling down. After moving these sample bags for further investigations, loose single sand grains, increasing in number with temperature, were detected in the bags for those samples heated above 500 °C. They indicate decreased cohesion of sand grains in the respective sandstones. In case of Cotta type sandstone this effect was less developed than for Posta type sandstone.

All baluster and cylinder specimens exposed for heating in the fire container show macroscopically visible cracks (Fig. 6b). Moreover, they reveal heavy sooting on the surfaces. Discoloration of the sandstone or crack surfaces could not be detected by the naked eye.

The temperature curves for the small sandstone specimens (50 × 25 mm) are shown in Fig. 6c. They were heated at different temperature levels of 400, 500, 600, 700, 800 and 1,000 °C in the laboratory oven at the TU BAF. The set-point temperatures (dashed lines) and the actual temperatures measured (solid lines) show a good correlation. It is recognisable that the heating experiments in the laboratory oven are precisely reproducible.

In contrast, the temperature curves in the fire container at a height of 1.8 m (solid lines) illustrate that the temperature increased very fast (after approx. 3 minutes) to max. 900 °C (Fig. 6d). After a dwell time of about 15 minutes (= the time the wood crib takes to burn through), the temperature in the fire container decreased rapidly.

If one compares the slow and even heating in the laboratory oven to the dynamic heating in the fire container, the differences between both treatments become obvious. According to the temperature curves measured by the thermocouples in the experiment displayed in Fig. 6b, on the stone surfaces (dashed lines), even in one and the same experiment, the temperatures range between 400 and 600 °C (Fig. 6d). The maximum temperature of about 600 °C is reached after approx. 14 minutes. The heating of the air in the container is faster and reaches higher maximum temperatures than the stone surfaces. However, the stone surface is cooling down much slower than the 93surrounding air. The upper right diagram in Fig. 6d shows remarkable lower temperatures within the stone compared to the stone surfaces. The maximum temperature of about 230 °C is reached only after approx. 55 minutes, i. e. long after the rapid decrease of the temperature of the surrounding air. Although the absolute temperatures measured may differ between single experiments, the general patterns of temperature development in the air, on the stone surfaces, and within the stone are similar. That means that the direct fire impact results in very unequal spatial and temporal distribution of temperature in the specimen within a short time of heating. These differences in temperature may lead to material tension caused by different thermic dilatation between the outer and the inner parts of the objects, resulting in cracks (Gómez-Heras et al. 2009). Many authors refer to the transformation of α-quartz to β-quartz at around 573 °C and the related volume increase to explain deterioration and damage of quartz-rich building stones (e. g. Chakrabarti et al. 1996, Hajpál & Török 2004). In the presented example, heavy damages (cracks) occur, although this temperature is hardly reached on the sandstone surface (see Fig. 6d).

Figure 6: a) Small specimens (50 × 25 mm) of Posta and Cotta sandstone parallel (PS_P & CS_P) and normal (PS_N & CS_N) to bedding after heating at different temperature levels in the laboratory oven at the TU BAF, from left to right: 25, 400, 500, 600, 700, 800 and 1,000 °C b) significant cracks and heavy sooting on the Posta type sandstone cylinder after the fire test at the IBK in Heyrothsberge c) temperature curves of the small sandstone specimens (50 × 25 mm) heated in the laboratory oven at the TU BAF for different temperature levels d) temperature curves of the thermocouples in the fire container at a height of 1.8 m (solid lines) and at the stone surfaces (dashed lines). The small diagram shows the temperature curves of the thermocouples inside the Posta type sandstone cylinder.

In the oven-heated smaller specimens, tension due to temperature gradients does not occur due to 94slower, even heating. From this point of view, this kind of experiment does not reflect real, short-term fire scenarios on buildings. However, these testings give insight into effects of heat on mineral grains and intergranular matrix. In case of long-lasting fire events, these effects may additionally affect building stones and their material properties.