The cellulose fiber used in this study came from the effluent generated by the alkaline extraction and bleaching steps and donated by a pulp and paper company in the city of Itapeva, State of São Paulo, Brazil. The expanded polystyrene and the plasterboard were collected at construction sites in the city that resulted from plasterboard cuttings when they were installed in the buildings.

The gypsum recycling was performed by calcination at 180ºC according to the methodology described by Bardella and Camarini (17). According to Savi (18), recycled gypsum presents flexural strength values between 0.7 and 3.5 MPa, but studies show that the use of cellulose fiber and expanded polystyrene in the recycled gypsum matrix results in flexural strength values below 0.7 MPa (19). For this reason, a mixture of commercial gypsum and recycled gypsum was used in a ratio of 1:1.

Five types of gypsum matrix composites with different amounts of waste were produced, and two using only gypsum as a control sample, one with commercial gypsum (Control 1) and one with a mixture of commercial and recycled gypsum (Control 2). Table 1 shows the percentages for common gypsum (G), recycled gypsum (RG), expanded polystyrene (EPS) and cellulose fiber (CF). The amount of water for hydration of the gypsum used in the production of the composites was in a ratio of 1:1.

Table 1 shows that the percentage of gypsum used in the composite production was maintained at 40% (20% of common gypsum and 20% of recycled gypsum) in all studied traits, this quantity was implemented because preliminary studies showed that the use of a lower percentage of gypsum was not enough for the homogenization and binding of the mixture, resulting in brittle samples. Only the percentages of the expanded polystyrene (EPS) and cellulose fiber (CF) residues were varied to produce the composites.

For the production of the composite, the EPS underwent a crushing process using a MARCONI brand chipper, as the material was in the form of plates, which required particle reduction with dimensions ranging from 2 mm to 3 mm to allow mixing with other materials. The EPS was then manually mixed with the cellulose fiber, gypsum, and water, until a homogeneous paste was formed, which was then transferred to the molds and kept at room temperature for 6 days for drying.

For thermal conductivity tests, samples of 50 mm in diameter and 20 mm in thickness were molded as specified by the ASTM E1530 standard (20). And for resistance tests, samples of 300 x 400 x 15 mm dimensions were molded, following specifications of NBR 14715-2 (21). The tests were performed using the samples with the gypsum in a hardened state.

After 6 days of drying, with the gypsum in a hardened state, the density of the studied composites was determined, calculating the division of the sample mass by its volume.

The thermal conductivity test, according to ASTM E1530 (20), was performed using the Thermal Conductivity Tester DTC-300 (Discovery, USA) which operates with steady-state heat flow meters in descending heat flow. For the measurements, the samples (Figure 1a), in hardened state, were impregnated with high thermal conductivity paste, used to decrease the thermal resistance between the surface of the sample and the equipment (Figure 1b).

The thermal resistance of a material is defined as the capacity it has in opposing the heat flow due to a temperature difference. In other words, in a material heat transfer can occur by conduction, convection or thermal radiation.

In this work, the most relevant mechanism was heat conduction, governed by the Fourier Law, where the thermal conductivity (k) was equivalent to the amount of heat transmitted by a time interval through a thickness, in a normal direction to the surface area, due to temperature variation.

The bending and compression strength test was performed according to ABNT NBR 14715-2 (21), using the universal EMIC test machine, where the sample (Figure 2a) was placed on two parallel bases with 350 mm spacing. A load was applied in the central part of the sample at a speed of 250±10 N/min until failure of the element (Figure 2b) for the quantification of burst load and strength.

Mean values, the mean confidence intervals (at the 95% confidence level) and the Tukey test results of the density of each composite studied are shown in Figure 3.

From Figure 3, it can be observed that the reduction in density is more pronounced in the composites that have a higher presence of expanded polystyrene, where C2 and C4 presented decrease of approximately 47% in comparison with the control sample of commercial gypsum (Control 1) and 44% when compared with the sample of recycled gypsum (Control 2).

While the higher presence of cellulose fiber resulted in a lower density reduction, it can be seen that C3 and C5 showed a reduction of approximately 37% and 27%, respectively, in relation to the common gypsum sample (Control 1), and approximately 34% and 23%, respectively, in relation to the sample with recycled plaster (Control 2).

San-Antonio-Gonzalez et al. (12) found density values between 460 and 660 kg/m³ for plaster composites with EPS. But, previous studies indicate that the density is mainly controlled by the amount and size of the expanded polystyrene particle aggregated to the gypsum matrix because of air contained in its interior (22), as can be seen in the graph of the relationship between the percentage of EPS used and the density of the composite, shown in Figure 4.

The graph in Figure 4 shows the linear Equation [1] that represents the relationship between the amount of EPS used in the composite production and the final density of the material, allowing to estimate density values for other percentages of EPS.

Figure 5 shows a graph relating the average values of density and thermal conductivity of the composites studied, where it can be observed that the samples with the highest thermal conductivity values were the ones with the highest densities, showing that the density directly affects the capacity of thermal insulation of the material. It is also observed that the higher the CF content, the greater its thermal conductivity and mass density, and the lower its insulation capacity.

The density of the material, according to the Brazilian standard NBR 15220-2, is directly related to its thermal conductivity, since the lower the mass density, the lower the thermal conductivity, and, therefore, the better its insulation characteristics (23). The results of the present study indicate that there is a correlation between density and thermal conductivity in gypsum composites (24, 25), and, according to the results in Figure 5, this correlation was again satisfied.

These values enabled obtaining the linear equation of the relationship between density and thermal conductivity of the studied composites (Equation [2]), which allowed to estimate the thermal conductivity of the composite for the other densities obtained with different EPS percentages.

Table 2 presents some thermal conductivity values obtained for light gypsum composites with different residues in the studies found in the literature. It was observed that the compositions studied in this work presented results compatible or better for use as thermal insulation.

Comparing the values obtained in the study with the thermal conductivity values in the Brazilian standard NBR 15220-2 (23) for gypsum board of 0.35 W/mK, it was observed that the composite produced with gypsum matrix and residues presented better thermal insulation results in all the studied configurations, with a reduction of approximately 48% for C4 and C2 and 8.57% for C5 in the thermal conductivity values.

Figure 6 shows the mean values, the confidence intervals of the means (at the 95% confidence level) and the Tukey test results of the flexural and compression strength of each studied composite and of the control samples.

The greatest reduction in bending and compression strength was obtained in the composites containing a higher percentage of expanded polystyrene (C4 and C2), with a reduction of around 48% in relation to the sample Control 2. The best result was found in the composite with a higher percentage of fiber cellulose in the gypsum (C3 and C5), with a reduction of 31%.

The results show that the higher amount of CF improves the flexural and compression strength of the composite in comparison to the EPS. This reduction in mechanical performance when plastic waste was added was also observed in the works carried out by Macedo et al. (8), Serna et al. (9), San-António-Gonzalez et al. (11, 12), Alameda et al. (13) and Del Rio Merino et al. (15).

The EN 13279 (26, 27) and plaster manufacturers specify that to be considered a lightweight plaster composite, the material must meet the following minimum values: density ≤ 800 g/cm3, flexural strength ≥1 MPa, and compression strength ≥ 2 MPa.

As shown in Figure 6, the flexural and compression strength decreases when compared to reference samples of commercial gypsum and recycled gypsum, but these values remain above the minimum required by EN 13279-1 for light gypsum composites, therefore all the composites satisfy this requirement (26).

The data obtained allowed to analyze the relationship between density and flexural strength of the material (Figure 7), where increasing the amount of EPS in the composite composition, the density decreases and consequently its mechanical strength. The graph also shows the linear equation of the relationship between the density of the material and the flexural strength (Equation [3]).

The results of this relationship between the two properties are compatible with those found by Santa Cruz Astorqui et al. (28), where the authors concluded that there is an exponential relationship between the reduction of the composite density with the addition of EPS and its strength.