Evaluation of the physical and mechanical behaviour of rammed earth by incorporation of recycled glass

b o l e t í n d e l a s o c i e d a d e s p a ñ o l a d e c e r á m i c a y v i d r i o 6 4 (2 0 2 5) 100442 www.elsevier.es/bsecv Evaluation of the physical and mechanical behaviour of rammed earth by incorporation of recycled glass Jacinto Canivella, Juan Jesús Martin-del-Rioa, Mario Solísb, José Daniel Rodríguez-Mariscalb, Vicente Flores-Alésa,∗, Francisco Pontigac aDepartment of Architectural Construction II, Universidad de Sevilla, Av. Reina Mercedes 4, 41012 Sevilla, Spain bDepartment of Mechanics of Continuous Media and Theory of Structures, Universidad de Sevilla, Camino Descubrimientos, s/n – Isla Cartuja, 41092 Sevilla, Spain cDepartment of Applied Physics II, Universidad de Sevilla, Av. Reina Mercedes 4, 41012 Sevilla, Spain a r t i c l e i n f o Article history: Received 5 March 2025 Accepted 2 May 2025 Available online 2 June 2025 Keywords: Rammed earth Compressive strength Non-destructive testing Aggregate substitution Crushed glass a b s t r a c t Compacted soil walls provide environmental benefits due to their low impact and embodied energy. Although their mechanical strength is lower than other materials, they meet safety requirements. Current trends promote circular economy solutions, such as waste reuse. This study assesses recycled glass in limeand cement-stabilized rammed earth as a replacement for natural sand to enhance waste management and soil properties. Ultrasonic inspection confirms increased compressive strength and density, especially in cement-stabilized mixtures. Statistical analysis reveals a direct correlation between crushed glass content and improved properties. The optimal replacement rate is 75% for lime and 100% for cement. The superior performance in cement-stabilized samples is due to a pozzolanic reaction absent in lime mixtures. Both binders show significant reductions in thermal conductivity, improving energy efficiency without compromising structural integrity. Ultrasound proves reliable for predicting compressive strength and stiffness, supporting the viability of this approach. The incorporation of recycled glass in compacted soil offers a sustainable construction alternative, balancing environmental benefits with enhanced mechanical and thermal performance. © 2025 The Authors. Published by Elsevier Espa˜ na, S.L.U. on behalf of SECV. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/ by-nc-nd/4.0/). ∗Corresponding author. E-mail address: vfl[email protected] (V. Flores-Alés). https://doi.org/10.1016/j.bsecv.2025.100442 0366-3175/© 2025 The Authors. Published by Elsevier Espa˜ na, S.L.U. on behalf of SECV. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). 2 b o l e t í n d e l a s o c i e d a d e s p a ñ o l a d e c e r á m i c a y v i d r i o 6 4 (2 0 2 5) 100442 Evaluación del comportamiento físico y mecánico de la tierra compactada con incorporación de vidrio reciclado Palabras clave: Tierra apisonada Resistencia a la compresión Ensayos no destructivos Sustitución de áridos Vidrio triturado r e s u m e n Los muros de suelo compactado ofrecen beneficios ambientales por su bajo impacto y energía incorporada. Aunque su resistencia mecánica es menor que otros materiales, cumplen con los requisitos de seguridad. Las tendencias actuales fomentan soluciones alineadas con la economía circular, como la reutilización de residuos. Este estudio evalúa el uso de vidrio reciclado en tierra compactada estabilizada con cal y cemento como sustituto de la arena natural para mejorar la gestión de residuos y las propiedades del suelo. La inspección ultrasónica confirma un aumento en la resistencia a compresión y la densidad, especialmente en mezclas estabilizadas con cemento. El análisis estadístico revela una correlación directa entre el contenido de vidrio triturado y la mejora de propiedades. La tasa óptima de sustitución es 75% para cal y 100% para cemento. El mejor desempe˜ no en mezclas con cemento se debe a una reacción puzolánica ausente en las de cal. Ambos aglomerantes reducen significativamente la conductividad térmica, mejorando la eficiencia energética sin afectar la integridad estructural. El ultrasonido resulta fiable para predecir la resistencia y rigidez, validando esta estrategia. La integración de vidrio reciclado en suelo compactado es una alternativa sostenible, combinando beneficios ambientales con mejoras mecánicas y térmicas. © 2025 Los Autores. Publicado por Elsevier Espa˜ na, S.L.U. en nombre de SECV. Este es un art´ ıculo Open Access bajo la CC BY-NC-ND licencia (http://creativecommons.org/licencias/ by-nc-nd/4.0/). Introduction The construction with soil may seem a simple technology [1], but these techniques have been developed and improved to provide better process performance. It is also a cleaner, healthier and less energy-consuming process than others, such as brick making, in which a firing phase is necessary. The techniques differ by the application and handling of the material. Two large groups can be distinguished, one of them in which the construction with earth supports a load and the other where the soil is only a filler. In the first group are techniques such as rammed earth, adobe, cob and compressed earth block (CEB). The CEB is similar to adobe, but in this case, the soil is subjected to high static pressure, which increases the compressive strength of the material. They are building blocks made from a mixture of soil, sand and clay, and may also contain lime or cement as a stabilizer. Since they are built-on-site the technical requirements are low in comparison to CEBs. Moreover, rammed earth walls are made by dynamic pressure with low energy per impact, hence the formwork does not need to withstand extreme forces. In both cases they share similar physical–mechanical properties. The earthen walls have an advantage providing a high thermal insulation capacity [2,3], they regulate the environment of the house in a natural way. The CEB and rammed earth are not fired, so they preserve the original properties of the earth walls, regulate humidity and accumulate heat. This compacting technique can also be carried out in situ, whenever the construction site allows it, by means of a compression machine. To produce the CEB and rammed earth, only about 1% of the energy required to make a conventional brick is needed, with minimal CO2emissions. Another advantage is that compressed earth walls maintain a constant relative humidity of around 50%, requiring less energy to heat them than in a traditional building [4]. The incorporation of aggregates from waste is an alternative that has been investigated with the aim of improving the properties of the final product and increasing its sustainability [5–7]. The possibility of incorporating crushed glass is a simple and clean option that, according to previous research, will improve the thermal insulation capacity [8]. Waste glass has traditionally been used in compacted soil, as a substitute for sand in concrete [9] and geotechnical products, mostly as a filler material, as a plasticity modifier [10], although in those cases where an alkaline activator is incorporated, such waste can also develop chemical stabilisation capacity depending on its particle size [11–13]. Other authors have reported reductions in thermal conductivity of around 38% for lightweight concretes with the incorporation of 45% glass [14]. The mean value of the thermal conductivity of rammed earth wall can be defined in the range 0.5–1.7 W/mK [15], which indicates a relatively poor thermal performance of rammed earth walls. In relation to thermal conductivity, the thermal conductivity coefficient of glass from domestic sources is estimated at an average value of 0.8 W/mK, although this can vary depending on the characteristics and origin of the material [8], so its incorporation into rammed earth elements contributes to the thermal conductivity of these being at the lower end of the range [16,17]. The incorporation of glass in compressed earth walls has also been studied in order to improve particle stabilisation and to obtain an increase in mechanical capabilities [18]. For example, powdered glass has been used by an alkaline activation process [6,11] to achieve substantial improvements in b o l e t í n d e l a s o c i e d a d e s p a ñ o l a d e c e r á m i c a y v i d r i o 6 4 (2 0 2 5) 100442 3 Table 1 – Chemical analysis of the AFN clay, according to manufacturer data-sheet and Atterberg limits. Chemical analysis (%) Plasticity SiO2Al2O3Fe2O3TiO2CaO MgO Na2O K2O MnO L.O.I. L.L. P.I. 57.59 17.63 6.38 0.39 3.40 2.14 0.18 3.16 0.08 8.75 34 15 L.O.I.: lost on ignition; L.L.: liquid limit; P.L.: plasticity index. Table 2 – Mineralogical composition of the AFN clay. Minerals % by weight Orthoclase (K-feldspar) 5 Quartz 39 Albite (Na-feldspar) >1 Hematite 2 Calcite 4 Muscovite 37 Chlorite 13 compressive strength and durability. Alkaline activation has also been used to improve the consolidation of earth mortars for restoration [19] with valid results for low alkaline concentrations. Alkaline activation has also been considered for rammed earth [20] and for CEB [21] but without including the incorporation of recycled or powdered glass [22]. In both cases, it is highlighted that this type of chemical reaction provides a better bonding matrix and therefore a higher mechanical performance. Since there is scarce research on the incorporation of recycled glass in construction techniques using compacted earth, the aim of this research is the analysis of the physical–mechanical performance of incorporating crushed glass in stabilised rammed earth. Physical and mechanical properties will be evaluated, depending on different percentages of substitution of the finer fractions of natural aggregate. The research aims to determine whether there are significant improvements in the aforementioned properties and, if so, what percentage of glass substitution would be optimal, assessing its advantages and disadvantages. Material and methods The soil used in this study was artificial manufactured from separated fractions so that a homogeneity is achieved between different batches. The soil is composed by a mix of gravel, coarse and fine sands and clay. The lime was purchased from a local company while the clay came from a specialised manufacturer (Sio-2®), who supplied it in powder form so that it could be easily mixed dry with all the aggregates. The chemical analyses of the clay AFN from Sio-2 are depicted in Table 1. The mineralogical composition of the clay is shown in Fig. 1 and its quantification in Table 2. Crushed glass was studied as substitute of sand, maintaining the established percentage of gravel. The crushed glass was provided by Ravsa S.A., a glass recycling company. The glass was crushed, by a mechanical grinding machine. Aggregate sizes larger than 4 mm were removed. A soil type was designed that was composed of 15% gravel, 50% coarse sand, 20% fine sand and 15% clay, which were drymixed prior to the preparation of the specimens. According to UNE-EN 9333-1 sieving method, these proportions comply with a particle size distribution of good compactness, since as can be seen in Fig. 2A, the curve corresponding to the soil has no discontinuities and remains close to its corresponding Fuller and its fineness modulus (4.29) is similar to that of its Fuller (4.79). The crushed glass that will replace the sand fractions is represented in Fig. 2B, where the similarity of both aggregates is observed. The experimental design considered the partial and total substitution with crushed glass of the sand fractions. As the intention is to analyse the behaviour of glass as a substitute for aggregate, 50, 75 and 100 percentages were sought in order to observe possible changes in behaviour. In addition, a reference dosage without glass was considered. This glass is a by-product from laminated pans of glass and was chemically characterized by means of X-ray fluorescence (XRF) in a Panalytical X-ray fluorescence spectrometer (AXIOS) with Rh tube for elemental solid sample analysis (Table 3). The analyses were carried out on pearl processed at 1200◦C using lithium metaborate and lithium tetraborate as fluxes in a ratio of 34/66. The flux/sample ratio used was 9/1. This soil was stabilized by Portland cement (C) and hydraulic lime HL5 (L), both in a percentage of 5% by mass of dried soil, being within the optimal range stated by several authors [23–25]. These conditions, as well as the names of each dosage, are listed in Table 4. The mixing moisture or optimum moisture content (OMC) is critical to achieve optimum compaction. According to previous research, the determination of this moisture content is usually performed by means of the UNE-103500 standard or modified Proctor test [26] or their equivalents according to ASTM D698-12 standard [27]. In this case, four Proctor tests have been prepared for two types of stabilization and for the cases of 100% glass and no glass substitution, shown in Fig. 3. In Fig. 3A and B, it can be seen that the moisture contents are very similar for all cases and are around 8%, except for the cement with glass, which is reduced to 7%. Since the differences are so small and considering the difficulty of precisely controlling this margin during manufacture, it was decided to work with an OMC equal to 8%. Once the starting parameters were defined, the manufacture and curing of the test specimens was developed. First, each soil and glass fraction were oven-dried for 24 h at less than 100◦C, so that the mineralogical components were not affected and to ensure that all mixtures started from the same moisture content before mixing. The aggregate fractions corresponding to each dosage were then dry-mixed, and the stabilizer (cement or lime) was added to achieve a uniform mixture. According to the dry weight produced and having previously checked the moisture content of the aggregates, the 4 b o l e t í n d e l a s o c i e d a d e s p a ñ o l a d e c e r á m i c a y v i d r i o 6 4 (2 0 2 5) 100442 Fig. 1 – X-ray diffractogram of the AFN clay. Fig. 2 – Particle size distribution of soil (A) and soil with processed glass (B). Table 3 – Chemical analysis of glass. Elements SiO2Na2O CaO MgO Al2O3K2O SO3Fe2O3TiO2MnO P2O5LOI % 72.21 13.05 9.98 3.46 0.76 0.34 0.3 0.26 0.07 0.03 0.01 0.46 D.L. 0.02 0.01 0.03 0.02 0.01 0.01 0.01 0.04 0.01 0.01 0.01 C.L. 0.03 0.02 0.05 0.03 0.02 0.02 0.02 0.06 0.02 0.04 0.02 Rel.E. 0.019 0.073 0.0100 0.029 0.011 0.038 0.137 0.023 0.106 0.050 0.026 Traces F Ba S P Zn Pb Cr Sr Zr Mn Sn Cs Cu Ppm 1211 1200 1076 144 84.8 66.6 64.5 59.5 57.1 52.1 32.2 27 26 D.L.: detection limit; C.L.: cuantification limit; Rel.E.: relative error. b o l e t í n d e l a s o c i e d a d e s p a ñ o l a d e c e r á m i c a y v i d r i o 6 4 (2 0 2 5) 100442 5 Table 4 – Distribution and description of each dosage for the experimental phase (-R, stands for reference group). Batch Number samples Stabilization Glass C-R 6 Cement (5%) – C-50 6 Cement (5%) 50% C-75 6 Cement (5%) 75% C-100 6 Cement (5%) 100% L-R 6 Lime (5%) – L-50 5 Lime (5%) 50% L-75 6 Lime (5%) 75% L-100 6 Lime (5%) 100% Fig. 3 – Optimum moisture content and maximum density for cement (A) and lime (B). Fig. 4 – Preparation of samples from prismatic shape moulds. 6 b o l e t í n d e l a s o c i e d a d e s p a ñ o l a d e c e r á m i c a y v i d r i o 6 4 (2 0 2 5) 100442 necessary water was added until the OMC was reached. The mixing was carried out in a pan concrete mixer that allows working adequately with low water–stabilizer ratios and drier textures. The mixture was then compacted in the moulds in four layers of about 5 cm thickness. Once the first 3 layers had been compacted, a plastic film was applied before the last layer was compacted. In this way, a 15 × 15 × 15 cm test specimen, called A, and another 15 × 15 × 5 cm specimen with the same compaction, called B, could be obtained for the tests (Fig. 4). In order to maintain uniformity in the rate of compaction of all the specimens, the procedure developed by the authors [28] was followed, but adapted to the use of a jackhammer, for which the methodology described by [24,29] is taken as a reference. Basically, it consists of establishing a relationship between the specific compaction energy of the standard Proctor test and that of the mechanical tools used. For this purpose, the hammer manufacturer’s data were considered, with the number of blows (b) per minute (nb) being 26 b/min, according to an intermediate speed configuration, and the energy per impact being j = 1 J. By equating the specific compaction energies from the Proctor and the experiment, Eq. (1) can be obtained, which describes the time required to compact a 5 cm layer of material with the hammer described. In our case tmin = 14.26 s, so the compaction time was controlled at 15 s per layer, during which the hammer, which had a compaction surface of 5 × 5 cm, covered the entire surface of 15 × 15 cm of the test specimen. tmin =Vm VOMC nOMC nb m × g × h j, (1) Vmbeing the volume of the specimen layer (m3), VOMC the volume of the standard Proctor test layer (m3), nOMC the number of blows per Proctor layer – which is 26 blows, nbis the number of blows per minute of the hammer, m is the mass of the Proctor hammer (2.5 kg), g is the acceleration of gravity in m s−2, h is the height at which the Proctor hammer falls (0.305 m) and j is the energy per blow of the hammer (Joules). After finishing three successive layers, which formed the type A test sample, before pouring the mixture, the plastic film was placed to separate the fourth 5 cm layer, thus defining the type B test sample. After demoulding, the samples were left to dry during 28 days under the same environmental conditions (20 ± 2◦C and 65 ± 5% relative humidity). The tests carried out on each type of specimen are described below. Specimens B were intended for tests to determine the bulk density and open porosity, by means of a water saturation method in vacuum, following the procedure provided in UNEEN-1936 standard [30]. The thermal conductivities of the samples were determined by using a test set-up manufactured by PHYWE Systeme GmbH & Co. KG [31]. This same set-up has been used by several researchers to measure the thermal conductivity of concretes [32] and earth building materials [33]. The equipment consists of a 40 cm side house, thermally insulated, which contains a heat source in its interior. The side walls of the house have 21 cm square openings, and the samples to be tested can be fixed from the inside against the openings using tensioning screws. Since the frontal area of our samples (type B specimens, 15 × 15 cm2) was smaller than the apertures on the side walls, the samples were mounted in a 4 cm thick polyurethane frame. During the experiments, type K thermocouples were used to measure the air temperature in the laboratory (Tair), the temperature at the center of the external surface of the sample (Tout) and the temperature at the center of the internal surface of the sample (Tin). A tiny amount of thermal grease (HY710, Shenzhen Halnziye Electronics Co.) was applied at the contact points between the thermocouple tips and the solid surfaces to improve the thermal contact. All temperatures were recorded over time using a high-resolution data logger (TC-08, Pico Technology) connected to a PC, until steady-state conditions were reached (6–8 h). At that moment, the heat flux through the sample per unit area q (W/m2) can be obtained as q =Tout − Tair Rs =Tin − Tout e/(2) where Rsis the surface thermal resistance of the air boundary layer next to the sample, and e and  are the thickness and the thermal conductivity of the sample, respectively. Here, according to the standard ISO 6946:2017 [34], the conventional value Rs= 0.13 m2K/W for horizontal heat flow through a plane surface will be assumed. The thermal conductivity of the sample can then be obtained as,  =e Rs ×Tout − Tair Tin − Tout (3) Ultrasonic pulse velocity (UPV) tests were performed on A samples with a Pundit Lab system from Proceq company, equipped with two 54 kHz piezoelectric sensors. The procedures established in the UNE-EN 12504-4 standard were followed in this testing [35]. Since ten readings are taken at each location, a variability range of ±2% from the mean value was therefore taken. After this check, some atypical values were discarded, so that the average of all the filtered readings in each UPV direction and location studied was obtained. This criterion was taken into account to determine the ultrasonic pulse velocities UPV-X and UPV-Y, for those directions perpendicular to the compaction direction, and UPV-Z for the compaction direction. In addition, three heights were differentiated in the X–Y plane, depending on whether the transducers were located at the base (B), in the middle (M) or at the top (T) of specimen A (Fig. 5). Measurements at 14 and 28 days were taken in order observe and check the evolution and certain trends [28]. However, the mechanical test was only possible to carry out at 28 days, hence lectures of UPV at 14 days will serve as a simple checking of the internal evolution of the samples. Table 5 describes the list of UPV variables initially considered in this research. The compressive strength of A-type specimens was obtained after curing for 28 days. The compression tests were carried out using a universal monoaxial testing machine. The loading history was established considering UNEEN 14580:2006 Standard [36]. The amplitudes of the loading–unloading processes were defined by load levels, but they were displacement controlled at 1 mm/min rate. The UPV measurements in the cement and lime samples, and the corresponding 50%, 75% and 100% subgroups, were used to determine the dynamic modulus of elasticity “MOE” b o l e t í n d e l a s o c i e d a d e s p a ñ o l a d e c e r á m i c a y v i d r i o 6 4 (2 0 2 5) 100442 7 Fig. 5 – Distribution of location of sensors for UPV and A-type specimens. Table 5 – Description of the different UPV readings taken in accordance with location, age, and direction. UPV XT14 UPV YT14 UPV XM14 UPV YM14 UPV XB14 UPV YB14 UPV ZM14 UPV XT28 UPV YT28 UPV XM28 UPV YM28 UPV XB28 UPV YB28 UPV ZM28 Location Top • • • • Middle • • • • • • Bottom • • • • Age (days) 14 14 14 14 14 14 14 28 28 28 28 28 28 28 Axis X Y X Y X Y Z X Y X Y X Y Z E (MPa) according to Eqs. (4) and (5), which depend on UPV v (m/s) and density  (kg/m3). The relationship between UPV and MOE in rammed earth materials has been validated by previous research [37]: E =pv2 K, (4) K =1 − v (1 + v) (1 − 2v), (5) where v is Poisson’s ratio, whose value was set at 0.35 as proposed by other authors [37,38]. Finally, a statistical analysis of the results (independent samples T-Student test) is carried out to establish the possible existence of statistically significant differences in the physical and mechanical values between the groups of reference specimens and those corresponding to the different degrees of glass substitution. If the test is positive, it could be affirmed that the substitution of sand by glass has a statistically decisive influence on the analysed property. Results and discussion Physical results Table 5 shows the descriptive statistics of the bulk density and open porosity values obtained for the different mixes. Except for the case of porosity for the cement-based mixes, the reduced standard deviation and coefficient of variation values describe a representative and uniform data set. It can also be stated that the average values are as expected for this type of material [24,39], the densities being similar to those established by the Proctor test (Fig. 2) for the dosages without glass and 100% replacement. Fig. 6 shows that in the case of cement, the median density increases slightly with increasing the degree of glass substitution, reaching 2.01 g/cm3. Consequently, the porosity decreases to a minimum of 21.18%, which defines a low porosity wall, in line with the minimum range established for both new [40] and historical [41] compacted soil-based materials. In contrast to this trend, the behaviour of the lime- 8 b o l e t í n d e l a s o c i e d a d e s p a ñ o l a d e c e r á m i c a y v i d r i o 6 4 (2 0 2 5) 100442 Fig. 6 – Representation of dry density (A) and open porosity (B) according to the type of stabilization method and the percentage of glass. Table 6 – Statistical results of physical properties of samples. N Density (g/cm3) Porosity (%) AV SD VC AV SD VC C-R 5 1.96 0.08 0.04 23.50 2.87 0.12 C-50 5 1.96 0.04 0.02 23.79 2.23 0.09 C-75 5 1.99 0.08 0.04 24.23 2.21 0.09 C-100 5 2.01 0.03 0.02 21.18 2.40 0.11 L-R 6 1.91 0.04 0.02 26.72 0.98 0.04 L-50 5 1.92 0.03 0.02 25.77 0.99 0.04 L-75 4 1.85 0.03 0.01 28.86 1.05 0.04 L-100 6 1.82 0.02 0.01 27.56 0.86 0.03 N: number of data; AV: average; SD: standard deviation; VC: variation coefficient. based specimens shows a progressive reduction in density as a higher percentage of glass replacement is used, down to a minimum of 1.82 g/cm3. The porosity, however, presents the minimum at the 50% substitution (L-50: 25.77%), being in any case always higher than those using cement. From this dosage of glass onwards, the samples increase their porosity until they reach values of more than 30% for L-75. Therefore, with regard to the physical behaviour, there seems to be a differential response depending on the stabilizer used, and the degree of glass substitution induces, from a descriptive point of view, an influence on density and porosity (Table 6). The statistical analysis on the physical properties shows that when cement is considered, there is no significant variation in density and porosity for any of the glass substitutions (C-50, C-75 and C-100) with respect to the reference (C-R). However, when lime is used, there are significant variations in density for L-75 [t(8) = 26.127, p = .000] and L-100 [t(10) = 4.6989, p = .0008] and in porosity for L-75 [t(8) = −3.4699, p = .008]. This means that, when cement is used, it cannot be statistically proven that the substitution of glass has a decisive influence on the physical properties, although, as shown in Fig. 5, there is a slight progressive increase in the density averages and consequent reductions in porosity. When crushed glass is used, pozzolanic reactions can occur in the presence of the calcium hydroxide of the cement, which generates new compounds (hydrated calcium silicates) that are deposited in the pores of the material [42]. When using lime, as there are more pronounced variations at higher glass substitutions (75% and 100%), it is statistically possible to state that glass can decisively alter the physical properties. Mechanical results Table 7 shows the descriptive statistics of the simple compressive strength values at 28 days obtained for the different percentages of glass substitution, according to each type of binder. In all cases, the mechanical strength is higher for cement, as has been reported in other studies [43], with the difference between the two being of the order of 100% and reaching more than 200% when the glass substitution is 100%. In general, the dispersion of results is as expected for this type of heterogeneous materials, with CVs between 15 and 25%, except in the case of C-75 and C-100, which rise to 30% [28,44,45], while for lime (Fig. 7A) the dispersion is smaller, with the percentiles being more clustered around the mean of each group. Regarding the use of lime, it is observed (Fig. 7) that there is a maximum UCS in the substitution L-75, with an average of 1.83 MPa, the standard deviation and the coefficient of variation being lower than in the case of cement. This b o l e t í n d e l a s o c i e d a d e s p a ñ o l a d e c e r á m i c a y v i d r i o 6 4 (2 0 2 5) 100442 9 Table 7 – Statistical results of unconfined compression strength of samples. N UCS (MPa) AV SD VC  % C-R 6 2.19 0.50 0.23 – C-50 6 3.00 0.69 0.23 – C-75 6 3.90 1.20 0.31 – C-100 6 4.07 1.23 0.30 – L-R 6 0.78 0.30 0.38 109 L-50 5 1.61 0.39 0.24 86 L-75 6 1.83 0.25 0.14 113 L-100 6 1.31 0.21 0.16 210 N: number of data; AV: average m/s; SD: standard deviation; VC: variation coefficient;  %: percentage difference of mean values between cement and lime samples. Fig. 7 – Representation of UCS according to the type of stabilization method and the percentage of glass (A). Relation between UCS and density according to the type of chemical stabilizer (B). trend is not followed by cement, which shows a progressive increase in mechanical strength until complete substitution with glass. The better mechanical behaviour of the cement specimens can be related to the higher density and lower porosity (Figs. 6 and 7B). As other studies have reported, the fine particles of glass provide higher adhesion due to the pozzolanic effect [42,46,47]. The hypothesis of the formation of hydrated neosilicates by alkaline activation can also be considered. According to Harrison et al. [48] alkaline activation occurs with glass when its size is less than 20 ␮m, which improves the mechanical behaviour when the chemical reaction takes place. In addition, the use of crushed glass provides an aggregate with angular shapes that improves mechanical performance for aggregate compaction [49] and, together with the formation of the hydrated components of the cement, allows higher levels of mechanical strength to be achieved. According to the UCS statistical study, when considering the use of cement, there is significant variation for 75% glass substitution, C-75 [t(10) = −3.2205, p = .0092] and C-100 [t(10) = −3.4587, p = .0061]. For lime, the variation is more pronounced in all glass substitution grades, L-50 [t(9) = −3.9959, p = .0031], L-75 [t(10) = −6.6071, p = .00012] and L100 [t(10) = −3.5738, p = .0051]. Therefore, it can be stated that the partial substitution of aggregates by crushed glass has a direct relationship with the change in mechanical behaviour and that it is in lime where these changes, at the statistical level, are more appreciable. Thermal properties Fig. 8 shows the average values of the thermal conductivity (l) for each dosage of crushed glass, showing a gradual reduction with increasing glass content, except in the particular case of the use of lime for L-100, which remains significantly the same as for L-75. Thermal conductivity of dry-state, non-glass rammed earth without additions [50,51] is in the range of that reported in this research. On the other hand, glass has a reported thermal conductivity of 0.4–0.9 W/mK [52], depending on its composition, being significantly lower than that normally reported for the rammed earth (0.6–1.6 W/mK, according to Ref. [51]), it is possible that the use of glass reduces the value of 16 b o l e t í n d e l a s o c i e d a d e s p a ñ o l a d e c e r á m i c a y v i d r i o 6 4 (2 0 2 5) 100442 [53] V. Flores-Alés, J.M. Alducin-Ochoa, J.J. Martin-del-Rio, M. Torres-González, V. Jiménez-Bayarri, Physical–mechanical behaviour and transformations at high temperature in a cement mortar with waste glass as aggregate, J. Build. Eng. 29 (2020) 101158, http://dx.doi.org/10.1016/J.JOBE.2019.101158. [54] Q.B. Bui, J.C. Morel, Assessing the anisotropy of rammed earth, Constr. Build. Mater. 23 (2009) 3005–3011, http://dx.doi.org/10.1016/j.conbuildmat.2009.04.011. [55] J.D. Rodríguez-Mariscal, M. Zieli´ nska, M. Rucka, M. Solís, Anisotropy and compaction gradient assessment on rammed earth specimens through sonic tomography approach, Eng. Struct. 309 (2024) 118058, http://dx.doi.org/10.1016/j.engstruct.2024.118058. [56] J.R. Rosell, I.R. Cantalapiedra, Simple method of dynamic Young’s modulus determination in lime and cement mortars, Mat. Constr. 61 (2011) 39–48, http://dx.doi.org/10.3989/mc.2010.53509. [57] J.D. Rodríguez-Mariscal, J. Canivell, M. Solís, Evaluating the performance of sonic and ultrasonic tests for the inspection of rammed earth constructions, Constr. Build. Mater. 299 (2021) 123854, http://dx.doi.org/10.1016/j.conbuildmat.2021.123854.