Two types of volcanic tuffs which are found throughout the Island were selected: a white-dark located in Vegueta-Triana (Las Palmas de Gran Canaria) and a brown-coloured from San Francisco (Telde). Both of them were collected from different areas from several demolition sites in order to obtain a representative sample of each type of stone. Additionally, care was taken in their collection and selection in such a way that their physical integrity was maintained. Figure 3a and Figure 3b show their external appearance.

A Bruker D8 Advance diffractometer with Bragg geometry was used. The minerals were analyzed using the DIFRRACplus program and a Rietveld semiquantitive analysis with TOPAS software. As can be seen in Table 1, great differences in mineralogical composition were observed although albite, sanidite and chabazite were the main components of both types of stone.

Moreover, a chemical analysis in accordance with the UNE-EN 197:2000 European Standard had been carried out on a non-weathered sample of each tuff (Table 2). We can see a higher amount of silica and alumina in the white-dark tuff and a higher content of calcite in the brown tuff.

Given their heterogeneity, six samples were prepared for each test. For preparation, they were firstly reduced in size by hand followed by an adjustment using a wet-cut circular saw to achieve the dimensions of the samples. Afterwards, the samples were dried in a chamber at 40 ±5°C for two weeks until a constant weight was reached. The dry materials were then introduced into sealed polyethylene bags until testing.

Compression and flexural tests were carried out. For compression tests, samples of 100 mm × 100 mm × 100 mm dimension were prepared. 400 mm × 150 mm × 150 mm were used in the flexural tests. Both were carried out following the EN 1926:2006 (30) and EN 12372:2006 (31), European standards respectively.

Samples of 200 mm × 200 mm × 25 mm were cut for abrasion resistance tests. The importance of this test is based on its use in bases and pavements. It followed the EN 14617-4:2005 (32) European standard. Finally, Shore A, C and D hardnesses were applied in the materials (33).

The importance of the hydric performance of samples is clear in terms of weathering. It is especially interesting in these stones given the high relative humidity in the Islands as well as the aforementioned trend of eliminating renderings. In this case, the water absorption coefficient under atmospheric pressure followed by a desorption test developed by LNEC (34), together with water absorption by capillarity were carried out under the EN 13755:2008 (35) and EN 1925:1999 (36) European standards respectively. Furthermore, the bulk density of samples was determined under the EN 1936:2007 (37) European standard method, as was the open porosity. To assess the microstructure and pore network of the samples thin laminas were developed and observed using a petrographic Kyowa Bio-Pol 2 with a plane-polarized light (//N) on a thin lamina, 30 µm thick, of the tuff stones and adhered to a glass slide with epoxy resin. The images were obtained with camera adapted to the microscope: the Moticam 2300 3.0 M-pixel. The thin lamina was obtained from the more significant parts of both tuff stones.

Thermal conductivity and specific heat tests were carried out. As the samples are commonly exposed to the outdoor surface, it was considered that thermal emissivity could be of great interest in estimating the radiative performance of the constructive systems.

Manufacturer and Navacerrada et al described the thermal conductivity apparatus and its operation (38,39). It is based on the measurement of temperatures of both sides of samples placed between the hot source at 60°C and the laboratory conditions when compared to a reference material of known thermal conductivity. To guarantee the accurate measurements, the thermal flow is determined under steady-state conditions, by means of the heat conduction in the sample as well as the air - wall heat transfer (39). It commoly implies that measurements were taken at about 5–6 hours after the beginning. In this case, six samples of 200 mm × 200 mm × 25 mm were prepared for the thermal conductivity.

Given the importance of the water content in porous materials, a test with saturated samples was carried out. In this case, the samples were wrapped in polyethylene films to seal them. Given the evaporation of water during the test, the water content was determined by differences in weight when measurements were taken and after drying them in a chamber at 40±5°C up to a constant weight.

Specific heat was measured using a SDT Q600 unit Differential Scanning Calorimeter (TA Instruments). It followed the Instrument Technical steps for measuring the specific heat which consisted of a comparison among the baseline, DSC sample and DSC reference material. The latter was synthetic sapphire. The tests were ramped from 0°C to 130°C at 10°C/min, under a flow of N₂ (100 cm³/min) based on preliminary tests. For this test, the samples were cut into pieces of about 6 mm and 1 mm in thickness, with a total weight of about 50–60 mg. To minimize errors due to the water content in the samples, before testing they were dried for 48 hours. Furthermore, two sample platinum pans were used to control the loss of weight.

The hemispherical thermal emissivity was determined using a RD1 portable emissometer with an AE1 detector in the range of 3,000–30,000 nm. The device is based on the sample being heated to 82°C and a control by a differential thermopile with low and high emissivity areas. It works in such a way that it creates a linear correlation between the sample radiation and two reference materials according to the ASTM C1371-04a (40) standard. In this case, specimens of 100x100x10 mm³ were used.

Finally, the thermal performance of four types of constructive systems of façades was estimated by their simulation in a pseudotime-dependent state. It was carried out using Antesol V.6 software developed by Monroy (41). The constructive systems consisted of a masonry wall, 40 cm thick, with and without a lime rendering of 2 cm and a gypsum plaster of 2 cm to analyze the differences. The thermal resistance was estimated based on the result tests and the literature in the case of the renderings. Hence, the thermal resistance, in dry conditions, were 0.609 (m².K)/W and 0.633 (m².K)/W for white-dark tuff, and 0.654 (m².K)/W and 0.675 (m².K)/W for brown tuff with and without finishings, respectively. Given the importance of water content on the thermal resistance, a simulation with the stones in a wet state was carried out. In this case, the thermal resistance was 1.976 (m².K)/W and 1.764 (m².K)/W for the white-dark tuff and 1.376 (m².K)/W and 1.294 (m².K)/W for brown tuff with and without finishings, respectively.

In all cases, south orientation façades were stated. As it is supposed that the buildings were placed in an urban area, the albedo was fixed at 0.2. Furthermore, climatic data was provided by the Spanish Weather Forecast (42) for Gran Canarias and three types of climate were considerd at sea level (0–350 m), at 350–1000 m and over 1000 m according to the classification of the Spanish Thermal Code (43). The solar factor was assumed to be 0.8 on sunny days and 0.1 on cloudy days.

About the finishing properties, they were adjusted to the type of material. Roughness was 0.40 for both types of stones and 0.2 in for renderings and plasters. Outdoor absorptivity and emissivity was adjusted to the test values and the literature for the renders and plasters (0.90 of emissivity and 0.80 of reflectance). Indoor absorption was fixed at 0.3.

Complementary to the previous analysis, two buildings were monitored (Figures 4a and 4b). The equipment was a TESTO model 435. The thermal transmitance is estimated as a comparison between thermal fluxes from the outside and inside temperatures (44), based on the ISO 9869:2014 standard.

The monitoring was carried out during a summer season, when the relative humidity was usually higher and its effect on the performance of the building could be analyzed. Sensors were placed on the south façade, due to the comfort conditions outside and the need to achieve at least 15°C of difference for accurate measurements (45, 46). The tests lasted 7 days or 168 hours, taking into account that the thickness was under 90 cm, following the recommendations of previous researches (47). Both of the selected areas were exposed under the same conditions to solar radiation and wind as well as the lack of heat sources at the inner surfaces, that could affect the comparison. At the same time, both the indoor and outdoor temperatures were recorded using TESTO loggers.

Finally, the aim of this part of the research was to analyze the influence of water content on the thermal transmittance. It was evaluated, in the same building with the same construction system and external conditions, with the difference between the thermal transmittance on the ground floor, close to the pavement (20 cm from the ground), and the first floor.

Table 3 shows the results of the mechanical tests. The mechanical strength is less than other volcanic stones but is in accordance with that of some tuff stones (11, 13, 20). For instance Cappadocian tuff presented a mean value of 6.53 MPa and 2.16 MPa in dry and saturated conditions respectively (16), showing a reduction in the compressive strength when saturated. However, tuff properties are very different depending on their welding conditions and conservation, for instance western Mexico tuff studied by Pérez et al. showed a compressive resistance of 19.7 MPa when it is not altered and 14.9 when has been submitted to weathering. In the same work, but in the case of a yellow tuff, the resistance was 41.6 MPa when the tuff is fresh and reducing its resistance to 8.9 MPa when altered.

In this work, the compressive and tensile strength of the white-dark tuff was 22% and 76%, respectively, less than that of the brown tuff which is in relation to the physical properties in section 3.2. A significant variation in the properties has been found, especially in the white-dark tuff because of its porosity caused by the alteration of the matrix (16) in the presence of water. At the same time, the highest quartz content in the brown tuff justified its higher performance in terms of mechanical strength. It is in accordance with the statements of different authors (48). In addition, the variation was limited to 5% to get a representative value.

In addition, the average value of the flexural / compression strength ratio was 0.12 and 0.18 in the white-dark tuff and in the brown tuff, respectively. It implied that the elasticity modulus would probably be low (49), as it is common in traditional materials. Furthermore, it is in agreement with the strain for maximum strength. Tuff masonry showed a considerable plastic capacity as stated in the literature. Calderoni et al established the plastic capacity of the masonry of tuff stones, with a good transversal interlock, depending on the arrangement of the stones and the type of mortar (15)

The compression strength and hardness are also related. White-dark tuff shows the lowest values from 40 to 65 in C scale depending on the alteration of the matrix. Areas without alteration in the matrix were selected for the abrasion resistance test. This explained the limited differences of 5% between both stones. From the results it can be determined that both of the tuffs can be permitted on low-moderate uses, in accordance with UNE-EN 14617-4 (32). It means that they can be used horizontally, vertically indoors as well as vertical outdoors under protection.

Complementary to this, mechanical tests were carried out on the saturated samples. As was expected, both of them considerably reduced the compressive strength by 53.2% in the case of the white-dark tuff and 3% in the brown. This result is considered of great importance in this type of material due to its porosity.

The white-dark tuff showed a 16% lower bulk density than the brown tuff (Table 4). On the other hand, the latter showed less open porosity than white-dark tuff. The water absorption coefficients by capillarity (Figure 7 and Table 4) of white-dark tuff when compared to the brown one should be highlighted. It implies that the pore structure is different in both even though pores are smaller in the former.

In Figure 5, the images obtained on a thin lamina with the petrographic microscope at 400x are presented, obtained under a plane-polarized light (parallel nicols //N).

In Figure 5, we can observe the white-tuff //N at 400X, a salic trachytic phononolitic tuff with a vitreous matrix, with fragments of sanidine and a communicated porosity. The right-hand image shows the basalt-trachyte tendency of the brown-tuff. The brown tuff is a breccia tuff with a bigger welding rate of the matrix and a larger number of vitreous phases than the white tuff analyzed. Moreover, its porosity is less connected than that of the white-dark tuff, where areas without mass can be observed in the left-hand image.

Compared to other work found in the literature, porosity can vary from 38.29% in the Cappadocian samples (16) to 10.5% white-dark tuff and 27.4% yellow tuff from western Mexico (12). This performance is directly related to the higher examples of decay in the white-dark tuff compared to the brown, as well as the relationship between porosity and mechanical strength or hardness (50–52). It is of interest in these cases of external exposure to the elements which may easily cause damage and which reduces their strength drastically as previously mentioned, together with their durability (4), even more so by taking into account the salt crystallization probability due to the atmospheric conditions (53,54). In the Mexican case (12) the weathering suffered by tuff increased the porosity 3.1 times and 2.02 times in the cases of white and yellow tuff respectively. Weibern tuff (Germany) and Loseros tuff (Mexico) showed 37.1% and 7.06% respectively, in the study presented by Wedekind et al (55).

In Figures 6a and b, water absorption by capillarity of the tuff stones can be observed when exposed to capillarity water (Figure 6a) or immersed (Figure 6b). Differences between the brown and the white-dark are clear in terms of absorption; although they behaved in a similar way as regards the desorption process (Figure 6b). The latter requires almost two weeks to achieve a constant value compared to the 24 hours in the absorption process. This performance should be highlighted when water can rise through capillarity as highlighted by Veiga (34). The water absorption immersion is less than the Cappadocian tuff, which showed 38.29% by weight, although the time for absorption and desorption has not been obtained (16).

Hence, once again, the physical properties of the stones reveal the need to reduce the rate of infiltration of water and pollutants as well as their movement through the stone to control the mechanical performance (48) and reduce their weathering (56). Indeed, the wetness effect due to the relative humidity as well as the rain noticeably gave rise to a reduction in strength (15, 48) which is recovered when the stones dry. Because of this, extensive research has been carried out in recent decades to develop compatible treatments for stone-based protection (11, 12, 19).

In Table 5, a summary of the thermal properties of the analyzed stones is given. The results are of great interest since there is a lack of information regarding these properties in the literature, in spite of the fact that thermal conductivity is related to the bulk density and the porosity. However, a limited difference of 6% can be observed between both. It should be highlighted that both thermal conductivities are noticeably under the standard values considered for porous natural stones. A thermal conductivity of 0.55 W/(m².K) is given by the standard (57) which is in accordance with several research works which shown values from 0.52 to 1.79 W/(m.K) for higher bulk densities of about 2,000–2,100 kg/m3 (58–59). It showed that specific and detailed research should be carried out to characterize each building material clearly.

In Las Palmas de Gran Canaria, the annual relative humidity is 68% ranging from 65% in March to 71% in October (42). This high relative humidity is of great importance in this type of stone given the open porosity of the tuffs and their common exposure to the outside conditions. These circumstances imply an increase in the thermal transferences of the constructive systems due to the water absorption. In addition, a high risk of condensation should be taken into account for the indoor comfort conditions due to the high relative humidity when the temperature decreases.

As regards the former, Brodsky and Barker (58–59) stated that the thermal conductivity of wet samples doubles the values for the dry samples. Hence, in order to analyze the thermal conductivity dependence on moisture, a simplified model has been considered based on geometric mean (Equation 1). This simple theory was researched by Ashworth who compared it to five other theories and found that it was the most accurate in predicting the performance of stones (59).

Where λ is the thermal conductivity of air (λair), water (λwater) and rock (λrock); n is the porosity and wc is the water content (volume of water in the sample / bulk volume of the sample). It was assumed that the thermal conductivity of still air and still water was 0.026 W/(m.K) and 0.609 W/(m.K), respectively. Porosity values were 36% and 23% for white-dark tuff and brown tuff, respectively (Table 3). The model is based on the existence of two stages: pores filled with air when wc = 0 (dry conditions) which is used to calculate the thermal conductivity of the rock; the second stage with wc = n when the pores were filled with water, to predict the saturated value of the thermal conductivity. Based on this simplification, the thermal conductivity of the saturated rocks was 0.79 W/(m².K) and 0.55 W/(m².K) for the white-dark tuff and the brown tuff, respectively. It means that the thermal conductivity in the wet stage was more than 2 and 3 times higher than that in dry conditions, respectively.

The theoretical results were in agreement with the tests carried out in the wet state. In the latter, the methodology was the same as that of the aforementioned experimental tests with the difference that samples were firstly saturated in water and then wrapped in several layers of polyethylene film. Measurements were taken under steady-state conditions and water content was determined by differences in weight, when thermal conductivity was measured and the dry state followed at 40°C to achieve a constant weight. The thermal conductivities were 0.436 W/(m².K) and 0.475 W/(m².K) of white-dark tuff and brown tuff, respectively, at about 30% of water content. Thus, it implied that the thermal conductivity increased 70–75% with this water. Furthermore, the results confirmed the recommendation of using these stones under protected conditions in order to avoid, not only the acceleration of the stone decay but also the thermal conservation performance of the constructive systems.

As regards the specific heat capacity, as can be observed in Figure 7a, it varied with the temperature, mainly at low and high temperatures. Considering the colour of the façades and the latitude of the Canary Islands, the surface temperature may be estimated ranging from 10°C in the winter to 80°C in the summer. In this case, the specific heat capacity of samples can be stated at about 1.16 J/(kg.K) for the white-dark tuff and 1.30 J/(kg.K) for the brown (Figure 7b) with implies a 12% higher accumulation capacity in the latter in comparison to the previous one.

Once again, if the experimental results are compared with the standards, all of the samples showed a higher thermal accumulation capacity than it is supposed. In Spain, the standards stated for porous stones establish a specific heat capacity of 1 J/(kg.K) which implies that, for instance, the brown tuff showed a 30% higher accumulation capacity than is supposed. It is of great interest since the thermal performance of the constructive system (time lag, decrement factor and thermal inertia, among others) varied notably depending on the properties assigned.

As regards the thermal emissivity, it was similar to all the samples and to other common building materials. The differences between the samples are probably caused by the mineralogical components. In the case of hot climates, the high emissivity of the surface is of great importance since it avoids an overheating of the surfaces and reduces the thermal stresses of the building materials.

Finally, the pseudo time-dependent simulation showed that temperatures, time lag and thermal fluxes were reduced with the use of renderings. In Figure 8a, the time lag is about 1 or 2 hours higher in brown tuff with and without renderings when compared to white-dark tuff with the same composition. The differences are greater in mid and high altitude areas compared to coastal ones due to the most variable temperatures. Thermal fluxes (Figure 8b), in the winter, increased slightly 1–6% due to the colour of the façades considering that the renderings were white; in the summer, the thermal fluxes reduced up to 4 times in higher altitudes or 46% in coastal areas, both in cloudy summer days for the brown tuff or 30% to 38% for the white-dark tuff, respectively. Furthermore, taking into account the envisaged effects of climate change, strategies should focus on summer conditions. Then, the preservation of the earth-coloured renderings is recommended which improves the performance not only in the summer but also in winter conditions.

This recommendation should be highlighted taking into account the high relative humidity of the Canary Islands. Indeed, the thermal fluxes of the stones increased up to 3 times when the tuffs were saturated compared to the dry conditions in the winter (Figure 8a). Furthermore, the increase in the thermal transmittance in the constructive systems of the envelope leads to a higher risk of condensation (60) as well as a reduction in thermal comfort. In the summer (Figure 8b), the thermal fluxes also increase up to 2 times, especially on sunny days although the differences are lower than in the winter.

Hence, permeable and specialized renderings (61) considerably improved the energy conservation of the masonry due to the reduction in the thermal fluxes in both winter and summer conditions and its preservation from the humidity. Furthermore, durability increases due to their protection. The capillarity performance of the stones should be taken into account in such a way that the joint between the façade and sidewalk should be constructed correctly to avoid the infiltration of water while allowing the evaporation of the underground water.

In general, the thermal transmittance values of the ground floor were considerably higher than the that of the first floor. Under state conditions, thermal transmittance of the wall on the ground floor was 0.891 W/m².K compared to the 0.33 W/m².K of the first floor for the white-dark tuff; while, it was 0.64 W/m².K on the ground floor compared to 0.41 W/m².K on the first floor for the brown tuff. It is important to highlight the lack of precipitation in the last year and its influence on the surface water. In any case, the humidity coming from the ground provoked an increase of 2.7 times the thermal transmittance between both placements for the white-dark tuff and 1.6 times for the brown tuff.

These values confirmed the statement of Baker who insists on the better performance of traditional buildings under real circumstances than under simulation (3, 46). It was of great importance due to the influence of the thermal performance of the constructive systems on the energy demand of the building to achieve conditions of comfort (62).