4.1. Indirect Tensile Strength (ITS)
A total of 16 combinations of levels have been studied; 4 levels of %RA
2 levels of %S
2 levels of NA. In order to predict the variance of pure error three replicates were conducted for each combination.
Table 3 shows the results of the ITS and the ITSR for each sample studied. Furthermore, voids in mineral aggregates (VMA), air voids (Va) and voids filled with bitumen (VFB) have been included in it. Spanish PG-3 requirements are the following: voids in mineral aggregates must be greater or equal to 14%, air voids between 5% and 9% and no specification for voids filled with bitumen [
15]. All the results for these parameters are in perfect compliance with the previous values.
It was observed that the mixing times of the RA with the bitumen until reaching a coating of 100% are, at least, on the order of 1.5 times the mixing times required by the natural aggregates analyzed. In addition, the presence of RA makes the compaction of the hot bituminous mixtures more difficult. It is due to the greater roughness of the cement mortar adhered to the RA increases the internal friction between aggregate particles. Moreover, this difficult to compact the samples can be observed in
Table 3; when the VMA and Va voids tend to increase when the percentage of RA grows. This behavior is more prominent when the NA2 is used as the aggregate.
Table 3 also shows the ITS of the wet and dry samples (State).
Student’s
t-test was used to test individual regression coefficients on the proposed model (equation 2). According to this test, either the interaction between the percentage of water saturation and percentage of natural aggregate or the second order term related with recycled concrete aggregate is not statistically significant in the response of indirect tensile stress (ITS) of the specimen.
Table 4 shows the statistical
t-test results where H
0 states that
βj = 0 in a bilateral test. In
Table 4 “No” means that the null hypothesis must be rejected with a confidence level of 95% while “Yes” means the opposite.
Therefore, discarding the coefficients with no statistical significance from the full model and recalculating the regression coefficients based on the original variables, the ITS equation can be written as follows:
Table 5 summarizes the analysis of variance of the least squares problem. As can be seen, the
p-value associated with the regression F
0, the model is significant, i.e., there is at least one
βj which is different from zero. In addition, the R
2 is 0.86 which indicates that 14% of the total response variance is due to unexplained reasons.
The sum of squares due to lack of fit and pure error is also included in the same table. With these variance terms, a lack of fit test was performed and concluded that there is no evidence that the regression model does not fit adequately.
Taking into account that
x3 corresponds to the type of natural aggregate characterized by its quartz content, the expressions of ITS for schist and calcite dolomite aggregates are, respectively, the following:
A residual analysis was carried out to test the goodness of fit and by this way verify the suitability of the regression models of both natural aggregates. In
Figure 3 and
Figure 4, the expected residuals ranked by the rankit method against the observed ones assuming normal distribution with the same mean and variance of the latter are plotted for schist and calcite dolomite aggregates, respectively. It can be observed that the residues are normally distributed throughout the observed responses, which confirms the validity of the fit.
Figure 5 and
Figure 6 show the response surface model and the experimental data for NA1 and NA2 respectively. As was expected, the ITS of the NA1 is greater than that of the NA2. It is because of the siliceous nature of RA and NA1. Besides that, in both cases, the influence of the percentage of recycled concrete material upon the indirect tensile strength of the dry samples results positively. The model indicates that when calcite is used as natural aggregate the influence of RA is higher than in schist one. Then, as was also expected, the addition of RA favored more the weaker natural aggregate. In general, the positive effect of RA upon the ITS can be explained as a result of the roughness of the RA surface that increases the internal friction of the mixture. Summarizing, the higher the percentage of RA is, the higher the ITS is. That is, RA improves the mechanical properties of the mixture when used in percentages up to 60%.
The influence of the percentage of water saturation is the same for both cases, having a negative impact on the response. It was also an expected result; water could lead to the dissolution of some compounds of the RA, thus if humidity increases, the tensile strength goes down. The loss of adhesion between bitumen and the aggregates due to the presence of water can be explained thermodynamically. The adsorption process between water and aggregates are more favorable in terms of Gibbs free energy than between bitumen and aggregates [
21].
The interaction between RA and water affects less NA1, that has more percentage of quartz than NA2. This result could mean that the adsorption energy between the components of the mixture is higher in the case of NA1 than in that of NA2, which has less quantity of quartz. In order to establish this hypothesis, measures of surface energy (γ) both in bitumen and aggregates (natural and recycled) must be done. Beside of this, the moisture damage index (MDI) should be evaluated as the ratio between the total adhesive bond energy for the dry and wet sample [
22]. In addition, the siliceous nature of RA has a higher influence in the mixtures made with NA2 than in the mixtures made with NA1, because schist has a siliceous nature too, which reinforces the hydrophilic behavior of the mix [
23].
4.2. Indirect Tensile Strength Ratio (ITSR)
According to the
t-test, as shown in
Table 6, for the ITSR, only the percentage of RA is significant in the model.
Where βj are the coefficients of the coded variables and H0 represents the null hypothesis.
From the values of
Table 6, the relationship between ITSR and the percentage of RA has a decreasing behavior. The penetration of water in a bitumen film is not an easy problem to explain. Several mechanisms were proposed such as spontaneous emulsification due to an inverted emulsion W/O; pore pressure, because of the presence of water in the structure of the pore of the mixture and the aggregate characteristics. [
24].
Taking into account the last mechanism, the recycled material used in this research behaved as a hydrophilic material because of the high amount of silica in it. Therefore, the capacity of the water to displace the bitumen in the mixture is the main cause of stripping of the mixture in this case.
According to Equations (1) and (4), the maximum RA percentage to achieve an ITSR equal or higher than 80% (minimum required by the Spanish specifications) is 16.8% in the case of mixtures made with schist as natural aggregate. Similarly, in the case of mixtures made with calcite dolomite as natural aggregate, the maximum RA percentage to achieve the minimum ITSR required by the Spanish specifications is 11.4%.
The above percentages could be improved using several processes of transformation of RA to improve the behavior of the mixtures against water. Pasandín and Pérez cured in the oven the mixtures for 4 h after mixing and before compaction at 170 °C. This methodology improved the stripping performance [
25]. In addition, the same authors [
26] proposed to pre-coat the recycled concrete aggregate with bitumen emulsion in order to improve the moisture damage resistance of mixtures made with RA. Lee et al. proposed to pre-coat the RA with a slag cement paste [
27]. Zhu and Zhong coated the RA with a patented liquid silicone resin [
28].
Another possibility is with surfactants as antistripping agents. Most of them are cationic, although anionic can be used in some circumstances. Recently, one of the authors published research about a cationic non-commercial surfactant that it could be used as anti-strip additive [
29].