1. Introduction
2. Literature review
3. Material and testing
4. Parameters calculation
5. Data analysis
5.1 Effect of RAP on S(60s) and thermal stress
5.2 Effect of RAP on complex modulus
6. SUMMARY AND CONCLUSION
1. Introduction
Construction and environmental costs of asphalt pavements can be reduced when using valuable recyclable materials is considered. Different types of environmentally-friendly solutions were introduced in the past decades; this includes Reclaimed Asphalt Pavement (RAP), reclaimed Portland cement concrete, iron blast-furnace slag, fly ash and waste tire. Among those materials, RAP has seen a significant increase in interest and usage over the years (NCHRP, 2001; McNichol, 2005).
It is well known that RAP has been used in U.S. for more than three decades (NCHRP, 2001; McNichol, 2005; Hansen et al., 2011). According to a recent report by Federal Highway Administration (FHWA), almost 73 million tons of RAP are reclaimed and 84% of those are successfully used annually in asphalt pavement construction, making RAP the most valuable recycled material in the U.S. (Hansen et al., 2011). In Europe, approximately 50 million tons of RAP are produced every year (EAPA, 2013; Aggregate business, 2014).
Based on field observations and forensic researches, road authorities and local agencies have proposed a number of specifications on the use of RAP for asphalt pavement application (NCHRP, 2001; McNichol, 2005). For example, in Minnesota, the current specification limits the use of RAP in asphalt pavement up to 40%, based on the traffic level and asphalt binder grade (MNDOT, 2008). However, still limited understandings on the effect of recycled asphalt materials on the mechanical properties of asphalt mixtures are left.
In this paper, the effect of adding different amounts of RAP on low temperature properties of asphalt mixtures is experimentally investigated. Dynamic Shear Rheometer (DSR) (ASTM, 2009) and Bending Beam Rheometer (BBR) tests (AASHTO, 2012) are performed at low temperature on a set of six asphalt mixtures. DSR and BBR experimental data are then used to obtain complex dynamic modulus, creep stiffness and thermal stress. The influence of RAP on these parameters is then finally evaluated based on statistical analysis and visual comparisons.
2. Literature review
Several studies were performed to investigate the effect of RAP on asphalt mixtures. In this paper, some of the most relevant to the present research are considered. In the recent past, the effect of adding RAP in asphalt mixtures was investigated by McDaniel et al. (2000). The experimental results demonstrated that the increased viscosity and stiffness, associated to the presence of RAP, improved rutting resistance at high and intermediate temperatures; however, RAP negatively affected the relaxation properties and fracture resistance at low temperatures due to a significantly higher material brittleness. Similar results at high temperature were obtained from Lee et al. (1999).
In a different study (Li et al., 2008), dynamic modulus test and semi-circular bend (SCB) fracture test were used to investigate the effect of adding RAP to asphalt mixtures prepared with different mix designs. Higher dynamic modulus, E*, was observed for mixtures containing RAP independently of the recycled material source; in addition, considerable decrease in fracture energy, Gf, was found in RAP mixtures as expected.
McGraw et al. (2007, 2010) investigated the effect of adding RAP on low temperature properties of asphalt mixtures. Recycled mixtures exhibited higher moisture sensitivity, while the binder extracted from these mixtures showed higher low PG limits in comparison with the original virgin binder.
In a recent research (Cannone Falchetto et al., 2014) on asphalt mixture strength at low temperature, it was verified based on statistical tools and size effect analysis, that RAP does not affect the peak stress at failure. More conventional studies (Moon et al., 2014) have demonstrated a substantial effect of RAP on asphalt mixture creep stiffness. In one of this research (Pranshoo et al., 2013) highest creep stiffness values were observed at low temperature for lower RAP content, while an opposite trend was observed at higher temperature.
In addition, field performance of RAP mixture was also addressed in recent German research efforts with the objectives of implementing material selection recommendations for practitioners (Wistuba et al., 2012).
3. Material and testing
A set of 6 asphalt mixtures, prepared with a Plain (unmodified) PG64-28 binder and a polymer modified (1.1% Elvaloy + 0.3% PPA) PG64-34 binder (AASHTO, 2010), was used in the present study (see Table 1); binder content was the same for all mixtures and equal to 5.4% by weight. The virgin aggregate materials consisted of screen sand, lime sand and 0.5 in. (12.5 mm) kram clear stone. RAP was provided by Korea Expressway Corporation (KEC); the recycled material presented a Nominal Maximum Aggregate Size (NMAS) = 19 mm, and was used in the mix design up to a maximum content of 40%. Asphalt mixtures were compacted with the Super-pave Gyratory Compactor (SGC) and short-term aged before testing.
Table 1. Asphalt Mixtures
BBR creep tests were performed on thin asphalt mixture beams (102.0±5 mm × 12.7±0.5 mm × 6.25±0.5 mm) according to the method proposed by Marasteanu et al. (2009). Similarly to the current procedure for asphalt binders (AASHTO, 2012), the mid-span deflection, δ(t), of asphalt mixture beam was recorded and used to obtain creep stiffness S(t) and the relaxation parameter m-value. Since asphalt mixtures are much stiffer than binders, test duration was extended from 240s to 1000s and a load higher than the standard 980mN±50mN was imposed. It was recently found (Cannone Falchetto et al., 2013; Moon et al., 2014) that BBR asphalt mixture beams are larger than the Representative Volume Element (RVE) of the material when used for evaluating creep properties at low temperature. This was demonstrated based on digital image processing techniques and on the analysis of the asphalt mixture microstructure (Cannone Falchetto et al., 2013; Moon et al., 2014). This further supports the use of BBR equipment for asphalt mixture creep tests at low temperature (see Fig. 1).
Based on the low PG limit of the asphalt binder used to prepare the corresponding mixtures, two BBR testing temperatures were identified: low PG+10°C and low (PG+10°C) + 6°C. The experimental results were then used to generate relaxation modulus, E(t), master curves and to estimate thermal stresses and critical cracking temperature. Five replicates were tested at each temperature; hence, a total of ten specimens were used per mixture type (see Table 2)
Table 2. BBR Mixture Creep Testing
Mixture | Binder type | PG | Temperature (replicates) | Applied load |
1,2,3 | Plain | PG64-28 | -18°C (5), -12°C (5) | 6,000 mN |
4,5,6 | Modified | PG64-34 | -24°C (5), -18°C (5) | 6,000 mN |
DSR tests were used to determine the complex shear modulus, G*, on asphalt mixture prismatic specimens (49 mm in length, 12 mm in width and 9 mm in thickness) at low temperature (ASTM, 2009). Due to the limited torque power of the available DSR testing device, together with the higher stiffness of asphalt mixtures, the specimen thickness was reduced to 6.25 mm. Such a dimension was selected also to facilitate the specimens’ preparation, since they could be easily obtained from previously prepared BBR mixture beams used for creep test. In addition, in order to prevent specimens’ failure in the areas in contact with the DSR loading fixtures, two C-shaped steel plates (3.5 mm in width and 0.25 mm in thickness) were additionally attached to the two ends of the asphalt mixture prisms. Fig. 2 presents the DSR specimens together with the loading setup on testing equipment.
DSR mixture complex modulus tests were conducted at two different temperatures: PG+10°C and (PG+10°C) + 6°C, with angular frequency ranging from 0.1 to 10 Rad/sec. In the case of mixtures designed with modified binder (PG58-34), only one temperature equal to (PG+10°C) + 6°C = -18°C was selected due to the high stiffness of the material at lower testing temperatures. In addition, only two replicates were tested for each mixture type due to limited testing conditions (see Table 3).
Table 3. Summary of DSR Mixture Testing
Mixture | Binder type | PG | Temperature (Strain) |
1,2,3 | Plain | PG64-28 | -18°C (0.0008%), -12°C (0.001%) |
4,5,6 | Modified | PG64-34 | -18°C (0.0008%) |
4. Parameters calculation
BBR mixture creep stiffness, S(t), can be computed based on Bernouli-Euler beam theory as:
$$S(t)=\frac1{D(t)}=\frac\sigma{\varepsilon(t)}=\frac{P\cdot l^3}{4\cdot b\cdot h^3\cdot\delta(t)}$$ | (1) |
where S(t) is the flexural creep stiffness, D(t) is the creep compliance, σ is the maximum bending stress in the beam, ε(t) is the bending strain, P is the constant applied load, l is the length of specimen (=102 mm), b is the width of specimen (=12.7 mm), h is the height of specimen (=6.25 mm), δ(t) is the deflection at the mid-span of the beam, and t is time.
Hopkins and Hamming’s algorithm (1967) was used to convert creep compliance, D(t), to relaxation modulus, E(t) and, from E(t), thermal stress was finally derived. First, based on CAM model (Marasteanu and Anderson, 1999) relaxation modulus, E(t), master curves were generated as:
$$E(t)=E_g\cdot\left[1+\left(\frac1{t_c}\right)^v\right]^{-w/v}$$ | (2) |
where Egis the glassy modulus (assumed: 30~35GPa for mixtures and 3GPa for binders: Moon et al., 2014), and tc, v and w are fitting parameters. Then, thermal stress was calculated from the one dimensional hereditary integral as can be seen in Equation (3):
$$\sigma(\xi)=\int\limits_{-\infty}^\xi\frac{d\varepsilon(\xi')}{d\xi'}\cdot E(\xi-\xi')d\xi'=\int\limits_{-\infty}^t\frac{d(\alpha\triangle T)}{dt'}\cdot E(\xi(t)-\xi'(t))dt'$$ | (3) |
where 𝜀(𝛏′) is strain and x is the reduced time. Equation (4) can be solved numerically using Gaussian quadrature with 24 Gauss points (Moon et al., 2013). In this paper, thermal stresses were obtained for a range of temperatures from 22°C to -40°C assuming a cooling rate of 2°C/hour.
5. Data analysis
In this section, creep stiffness, S(60s), thermal stress at PG+10°C, σ(PG+10°C) and complex shear modulus, |G*| are statistically and graphically compared. S(60s) and thermal stress were evaluated through analysis of variance (ANOVA) (Cook and Weisberg, 1999) with a significance level of 0.05 (α=0.05). Given the limited number of replicates, the results of |G*| obtained from DSR mixture tests were only compared visually.
5.1 Effect of RAP on S(60s) and thermal stress
In order to investigate the effect of RAP on S(60s) and thermal stress, two statistical groups (Table 4) were set based on the asphalt binder used for preparing the corresponding mixture. Since mixtures 3 and 6 were designed with virgin aggregates (NMAS = 12.5 mm) they were set as control groups for the mixtures prepared with PG64-28 and PG64-34 binders, respectively. Fig. 3 and 4 provides the experimental results of of S(60s) and thermal stress.
Table 4. Statistical Experimental Design (Group 1: PG58-28 and Group 2: PG58-34)
Statistical analyses were performed to investigate the effect of RAP (0%, 25% and 40%), on S(60s) and thermal stress obtained at testing temperature of PG+10°C. Two-way ANOVA was selected for the analysis with the assumption of a linearity between factors (RAP and Air voids) and response. Normality and constant variance conditions were met by thermal stress, while, based on Box-Cox analysis, S(60s) was log- transformed to reduce the uneven distribution of residual errors (Cook and Weisberg, 1999). Significance level was set to α=0.05. Table 5 summarizes the ANOVA results for S(60s) and clearly indicates that mixtures containing RAP (25% and 40%) present significantly higher stiffness, S(60s) values compared to the corresponding control mixture. In addition, significant two-way interaction terms were observed for S(60s). Air voids content did not affect the stiffness of mixtures (plain, PG64-28, and modified, PG64-34).
Table 5. ANOVA Results of S(60s): Group 1 and 2
Analogously to the previous statistical analysis and based on asphalt binder type, two groups were identified (see Table 4) for investigating thermal stress. The ANOVA results are presented in Table 6.
Table 6. ANOVA Results of 𝜎(-18°C): Group 1 and 𝜎(-24°C): Group 2
Higher thermal stress results are associated to increased RAP content for both plain PG64-28 and modified PG64-34 mixtures. Multiple comparisons were also performed to evaluate any possible significant differences in creep stiffness and thermal stress values between mixture containing 25% and 40% RAP (see Table 7). Nevertheless, no significant differences in thermal stress were found for mixtures prepared with binder PG64-28. On the other hand, mixtures prepared with modified binder PG64-34 showed statistically significant differences in the response.
Table 7. Multiple comparisons of creep stiffness and thermal stress (Group 1&2)
From the statistical analyses results, it can be concluded that when a plain (stiffer) binder, PG64-28, is used to prepare recycled asphalt mixture, no significant differences in low temperature properties are observed when RAP content exceed 25% and up to 40%. However, when using a modified (softer) binder, PG64-34, there is a statistically significant decay in low temperature performance when increasing RAP from 25% to 40%.
5.2 Effect of RAP on complex modulus
In this section, DSR mixture tests (ASTM, 2009) were used to evaluate the effect of RAP on shear complex modulus, G*. Figs. 5 and 6 show the results of |G*| of asphalt mixtures prepared with binders PG64-28 and PG64-34 for different angular frequency (0.1~10 rad/sec).
In the case of mixtures prepared with plain binder (PG64-28), DSR tests were performed at two different temperatures: -18°C and -12°C. Therefore, |G*| master curve could be obtained using a shift factor, aT. However, for mixtures prepared with binder PG64-34, DSR tests could be performed only at one temperature, -18°C, due to device limitations, as mentioned in previous section. Hence, |G*| master curves could not be generated for this binder type.
Figs. 5 and 6 graphically and quantitatively indicate that higher values of |G*| are associated to higher RAP content. Smaller differences in |G*| could be observed for RAP content between 25% and 40 in both cases.
6. SUMMARY AND CONCLUSION
In this paper, the effect of adding Reclaimed Asphalt Pavement (RAP) on low temperature properties of asphalt mixtures was experimentally and statistically investigated. A total of six asphalt mixtures were prepared with different mix design obtained from the combination of RAP (0%, 25% and 40%), air voids (4%) and binder type (plain: PG64-28 and modified: PG64-34). Three point bending and torsion tests were performed on asphalt mixture specimens with Bending Beam Rheometer and Dynamic Shear Rheometer, respectively. Creep stiffness, thermal stress and complex shear modulus were computed and, then, statistically and graphically compared. Based on the performed analyses, the following conclusions were drawn:
∙ Significant increase in creep stiffness at 60 seconds were observed in mixture containing RAP. Nevertheless, no significant differences in these low temperature parameters were found between 25% and 40% RAP content when plain binder (PG64-28), was used in the mix design. However, considerable variation in creep stiffness at the same two RAP percentages were detected for mixtures prepared with modified binder (PG64-34).
∙ Higher thermal stresses were found for mixtures containing 25% and 40% RAP in comparison with mixture prepared with virgin materials. Similarly to the previous results for creep stiffness, no significant differences in thermal stresses could be observed for RAP content between 25% and 40% when plain (PG64-28) binder was used in the mixtures preparation.
∙ Increasing RAP content resulted in higher shear complex modulus. However, for mixture prepared with plain binder (PG58-28), the stiffening effect of recycled material on |G*| decreases between 25% and 40% content.
Based on the results of the present research, the current specification, which allows incorporating RAP up to 40% into asphalt mixtures, should be more asphalt binder and mix design specific. However, in order to provide a more comprehensive perspective on the on the use and content limits of RAP, additional experimental data and analysis are needed.
Even though significant differences were observed between virgin and RAP mixtures, no significantly different low temperature properties were measured in mixture prepared with plain binder (PG58-28) and containing RAP between 25% and 40%. This is not true for mixture prepared with modified binder (PG64-34). Therefore, it is hypothesized and higher RAP content can be incorporated into asphalt mixture prepared with stiffer binder (PG64-28), without significantly changing its low temperature properties.