Understanding The Performance of Modified Asphalt Binders in Mixtures: High-Temperature Characterization

Phase 1--Evaluation of High-Temperature Asphalt Binder Tests
Using Mixtures With Limestone and Diabase Aggregate

A. Background

This report supplements a Federal Highway Administration (FHWA) report titled Understanding the Performance of Modified Asphalt Binders in Mixtures: Permanent Deformation Using a Mixture With Diabase Aggregate (Publication No. FHWA-RD- 02-042).⁽¹⁾ The research findings in FHWA-RD-02-042 and in this report were obtained under a study that is partially funded through National Cooperative Highway Research Program (NCHRP) Project 90-07. The objective of the study is to determine if asphalt binder performance is captured by the Superpave asphalt binder specification developed under the 1987 through 1993 Strategic Highway Research Program, with an emphasis on evaluating the performances of mixtures containing polymer-modified asphalt binders with identical Superpave performance grades (PG's), but varied modification chemistries.⁽²⁾ Although identical PG's were desired, the high-temperature PG's of the polymer-modified asphalt binders ranged from 71 to 77 after rolling thin-film oven (RTFO) aging.

Superpave uses the parameter G*/sind to grade asphalt binders according to their resistance to rutting at high pavement temperatures. At high temperatures, rutting resistance should increase as G*/sind increases. The asphalt binder with the highest G*/sind should have the most resistance to rutting. During the study documented in FHWA-RD-02-042, it was found that G*/sind at 50°C  had a high correlation to mixture rutting resistance as measured by the cumulative permanent shear strains from repeated shear at constant height (RSCH) at 50°C .⁽¹⁾ RSCH is applied by the Superpave Shear Tester. The r² was 0.89, but the degree of correlation was highly dependent on dynamic shear rheometer (DSR) frequency. G*/sind at 70°C had a weak correlation to mixture rutting resistance as measured by the French Pavement Rutting Tester (French PRT) at 70°C .⁽¹⁾ The r² was 0.70, although it increased to 0.88 after removing the data for 1 of 11 asphalt binders.

The objective of the rutting study was to determine which asphalt binders provide high-temperature properties that do not agree with mixture rutting resistance.⁽¹⁾ This would indicate what types of modification provide properties that are, or are not, correctly captured by the current Superpave asphalt binder specification. In general, the number of discrepancies between G*/sind and mixture rutting resistance was low. The data indicated that the current Superpave asphalt binder specification and testing protocols are valid for most of the asphalt binders tested in the referenced study.⁽¹⁾

This report is divided into two phases. In phase 1, asphalt binders were combined with a limestone aggregate to verify the conclusions provided by a mixture with diabase aggregate. Phase 2 addresses recommendations that the diabase mixtures be retested using RSCH at a test temperature closer to the PG's of the asphalt binders.

B. Objective

The objective of phase 1 was to verify the findings provided by a mixture with diabase aggregate using a second mixture with limestone aggregate. For the diabase mixture, the high-temperature properties of the asphalt binders correlated to mixture rutting resistance. Furthermore, a change in high-temperature PG from 70 to 76 increased rutting resistance at 50°C based on RSCH and at 70°C based on the French PRT.

C. Materials

Eleven asphalt binders were tested. This included one air-blown asphalt and eight polymer-modified asphalt binders: (1) styrene-butadiene-styrene [SBS] Linear, (2) SBS Linear Grafted, (3) SBS Radial Grafted, (4) ethylene vinyl acetate [EVA], (5) EVA Grafted, (6) Elvaloy, (7) ethylene styrene interpolymer [ESI], and (8) chemically modified crumb rubber asphalt [CMCRA]. There were two control asphalt binders: an unmodified PG 70-22 and an unmodified PG 64-28. The polymer-modified asphalt binders include elastomeric and plastomeric modifiers. Grafting includes any mode of chemically reacting a polymer with an asphalt binder, for example, vulcanization. The target PG for the polymer-modified asphalt binders was PG 73-28. The PG 64-28 asphalt binder and a PG 52-34 asphalt binder from the same crude source were modified. The air-blown asphalt was originally the PG 52-34 asphalt binder.

Four of the 11 asphalt binders were chosen for use with the limestone aggregate: Elvaloy, EVA Grafted, SBS Linear, and PG 64-28. These binders were selected because they provided relatively high and low levels of cumulative permanent shear strain using the diabase aggregate. Additional information on the asphalt binders, and information on the aggregates and mixture designs are given elsewhere.^(1-2)

D. Tests

High-temperature asphalt binder properties were measured by a DSR after RTFO aging.⁽³⁾ Mixture rutting resistance was based on the cumulative permanent shear strains from RSCH and on the rut depths from the French PRT.^(4-5) All mixtures were subjected to 2 hours (h) of short-term oven aging (STOA) at 135°C . Specimens were tested approximately 48 h after compaction.

E. Cumulative Permanent Shear Strain

Cumulative permanent shear strain from RSCH was measured at 7.0-percent air voids, 50°C , and 5,000 cycles. The applied shear stress was 69 ±5 kilopascals (kPa). The loading time was 0.1 second (s) and the rest time was 0.6 s. Three replicate specimens were tested per mixture. Lower cumulative permanent shear strains indicate more resistance to rutting. This test mimics fast, heavy pavement loads.

Table 1 gives the cumulative permanent shear strains from RSCH for the four asphalt binders with both the limestone and diabase aggregates. Table 2 shows that the replicate data for the limestone mixture with SBS Linear had one high strain relative to the other two strains. This high strain was considered a potential outlier. Therefore, averages with and without the high strain were calculated for this mixture.

Aggregate type did not affect the cumulative permanent shear strains from RSCH at a 5-percent level of significance. Thus, rutting performance was independent of aggregate type. The shear strains of 38,600 micrometers per meter (mm/m) and 28,400 mm/m using PG 64-28 are not significantly different because of the high variability of the shear strains for the limestone mixture. (See table 2.)

Table 1 shows that high-temperature PG provided a correct ranking, and an increase in PG from 70 to 76 would significantly decrease cumulative permanent shear strain. (Linear regression analyses were not performed because the number of data points are too low.) The PG's agree with the shear strains better than the G-star divided by sine delta's (G*/sind 's) of the asphalt binders measured at the RSCH test temperature of 50°C and 10.0 radians per second (rad/s). The use of other frequencies did not provide a correct ranking. G*/sind 's at a relatively low frequency of 0.1 rad/s are included in table 1. Based on the shear strains from RSCH in table 1 and on the full set of data for the diabase mixture at 50°C , it was concluded that the G*/sind 's for Elvaloy are low. (The full set of data are given in phase 2 of this study and in reference 1.)

RSCH applies a rest period after each cycle of loading while the standard DSR test for asphalt binders does not. The rest period allows a mixture to recover time-dependent recoverable elastic deformations. Because the DSR test does not include a rest period, these deformations will be part of the permanent deformation. To determine whether this was the cause of the discrepancy for Elvaloy, the four asphalt binders were tested by a method titled "Determining the Rutting Resistance of Asphalt Binder Subjected to Repeated Creep (RC) Using a Dynamic Shear Rheometer (DSR)."⁽⁶⁾ This test applies repeated loads with rest periods like RSCH. Each loading cycle had a duration of 1.0 s followed by a 9.0-s rest period. The applied shear stress was 25.0 Pascals (Pa). This test was used to measure the cumulative permanent shear strain at 100 cycles. This means that the asphalt binders and the mixtures were evaluated by the same parameter. Table 1 shows that the test did not provide an improved ranking.

F. French PRT

The French PRT tests a slab for permanent deformation using a rubber tire inflated to 600 ±30 kPa.⁽⁵⁾ Each slab had a length of 500 millimeters (mm), a width of 180 mm, and a thickness of 50 mm. The applied load was 5000 ±50 Newtons (N) and the test temperature was 70°C . The air-void level was 7.0 percent. The test normally ends at 6,000 wheel passes, but it was continued to 20,000 wheel passes to determine if this would change the relative performances of the mixtures. Although the measurement from the French PRT is generally called the "rut depth," it is actually a percent rut depth, which is the deformation in millimeters times 100 divided by the slab thickness of 50 mm.

Table 1. DSR and RSCH data.

¹Cumulative permanent shear strain of the asphalt binder.

Table 2. Cumulative permanent shear strain from
RSCH for the mixtures with limestone aggregate.

¹Possible outlier.

The average rut depths from the French PRT at 6,000 wheel passes are given in table 3. The data for individual slabs are shown in table 4. The coefficients of variation (CV) in table 4 are remarkably low for testing only two specimens per mixture. Only 2 of 17 mixtures had a CV above 20.0 percent.

Table 3 shows that for each asphalt binder, the average rut depth was lower using the limestone aggregate compared to the diabase aggregate with 4.85-percent asphalt binder. The range in rut depth was also lower using limestone. The asphalt binder content for the diabase mixture was then reduced from 4.85 to 4.55 percent to determine whether the difference in rutting resistance was related to the volumetrics of the mixture. This reduction increased the air-void level after 75 gyratory revolutions from 3.2 percent to the typically used mixture design level of 4.0 percent. The asphalt binder content for the limestone mixture was based on a 4.0-percent air-void level, so it was not changed. Table 3 and figure 1 show that the rut depths for the diabase mixtures with the lower asphalt binder content are very close to the rut depths provided by the limestone mixtures. Figure 1 also shows that the slopes are roughly the same, which indicates that the high-temperature PG's had a similar effect on all three types of mixtures. The data at 20,000 wheel passes provided the same conclusions. These data are given in table 5.

Styrelf, AC-10, and AC-5 (PG 82-22, PG 64-22, and PG 58-34) asphalt binders, which were used in prior FHWA studies, were tested with the limestone aggregate to expand the range in high-temperature PG.^(1,5) These data are given at the bottom of table 3. The data for all seven asphalt binders with the limestone aggregate are shown in figure 2, along with all of the data for the diabase aggregate using a 4.85-percent asphalt binder. These data are given in reference 1. The addition of the three data points for the limestone aggregate provided a curvilinear relationship that is roughly flat above PG 70. Both relationships in figure 2 show that asphalt binders with a PG of 70 or greater led to rut depths that were lower than the maximum allowable rut depth of 10 percent. The data appear to validate the Superpave system.

G*/sind 's at 0.9 rad/s and 70°C were compared to the rut depths from the French PRT because this frequency represents the slow speed of the device.⁽⁵⁾ (See table 3.) Figure 3 shows that G*/sind provided curvilinear relationships like high-temperature PG. The data for Styrelf are not included in figure 3 because its G*/sind was very high (2360 Pa). The G*/sind for EVA at 0.9 rad/s is low based on the mixture with the diabase aggregate.

The slow speed of the French PRT does not mean that the results are valid for Superpave standing traffic loads. All small-scale wheel-tracking devices have slow speeds, but the protocols and pass/fail criteria for them are generally based on data from pavements where vehicle speed is variable and the average speed is much higher than the French PRT speed of 7 kilometers per hour (km/h). Even with this confounding factor, it seems reasonable in research studies to adjust the DSR frequency to match the slow speed of a wheel-tracking device. The relationships based on 0.9 rad/s in figure 3 are reasonably good, but so are the relationships based on 10 rad/s in figure 2. The data for these materials do not show that one frequency is more appropriate than the other. (Additional information concerning the appropriate DSR frequency is included in phase 2 of this report.)

The effect of increasing the PG from 70 to 76 is difficult to determine because of the empirical nature of the French PRT. Figure 2 appears to indicate that bumping the PG from 70 to 76 to account for slower traffic speeds or high equivalent single-axle loads (ESAL's) may not provide a more rut-resistant mixture. All asphalt binders with a PG of 70 or greater provided rut depths that passed the test. However, if it is assumed that the French methodology was developed for fast traffic speeds, then the severity of the test, or the pass/fail criterion, is not sufficient for slow traffic speeds. Therefore, no overall conclusion concerning a bump in PG can be made based on figure 2. In order to conclude that a bump in PG would never be beneficial, the French PRT would have to apply the severest conditions. Most likely, it does not apply the severest conditions, although it is not clear what the French methodology represents in terms of vehicle speed and ESAL's. Based on data collected in this and previous FHWA studies, the device may simulate pavement loadings that require, at most, one bump in PG. If the severity of the test is equivalent to one bump in PG, then the bumped grade needed to have both mixtures pass the test is PG 70 and the original grade would be PG 64. This bump may not improve the rutting performance of the limestone mixture, but it is needed for the diabase mixture.

Figure 3 shows that all mixtures passed the test at a G*/sind of 200 Pa or greater. This is lower than the G*/sind of 2200 Pa used by the Superpave asphalt binder specification after RTFO aging because a DSR frequency of 0.9 rad/s was used to represent the speed of the French PRT. Plots like figures 2 and 3 are useful for determining whether an asphalt binder parameter correlates to mixture rutting under a set of conditions that roughly mimic average pavement loadings for a certain class of highway, such as Interstate highways. However, they should not be used to determine pass/fail criteria for asphalt binder tests unless it is known what the wheel-tracking device represents in terms of vehicle speed and ESAL's, and a DSR frequency that matches the speed of the wheel-tracking device can be established.

G. Conclusions

• High-temperature PG agreed with mixture rutting resistance based on both the cumulative permanent shear strain from RSCH at 50°C and the rut depths from the French PRT at 70°C . The data provided no reason to change the Superpave high-temperature asphalt binder specification.
  
  
• An increase in high-temperature PG from 70 to 76 decreased the cumulative permanent shear strain at 50°C .
  
  
• All mixtures with a PG of 70 or greater passed the French PRT. Even so, it could not be concluded that an increase in PG from 70 to 76 would not be beneficial because of the empirical nature of the device and its pass/fail specification.
  
  
• The cumulative permanent shear strains from RSCH showed that the G*/sind for Elvaloy at 50°C was low. G*/sind underpredicted the relative rutting resistance provided by Elvaloy. However, Elvaloy's high-temperature PG agreed with the strains from RSCH. Furthermore, the asphalt binder properties of Elvaloy agreed with the rut depths from the French PRT at 70°C . The data provided no reason to change the Superpave asphalt binder specification.
  
  
• The French PRT indicated that the G*/sind for EVA at 0.9 rad/s and 70°C was low. This anomaly is discussed in phase 2 of this report.

Table 3. DSR data and French PRT rut depths at 6,000 wheel passes.

¹AC = Asphalt content by total mass of the mixture.

Table 4. Percent rut depth from the French PRT at 70°C and 6,000 wheel passes.

¹CV = Coefficient of Variation, percent = (standard deviation ÷ average)*100.

Figures 1. Figure 1. French PRT rut depth
at 70°C vs. high-temperature PG.

Figure 2. French PRT rut depth at 70°C vs.
high-temperature PG for all asphalt binders.

Table 5. DSR data and French PRT rut depths at 20,000 wheel passes.

¹AC = Asphalt content by total mass of the mixture.

Figure 3. French PRT rut depth at 70°C
vs. G*/sind at 70oC for all asphalt binders.

back to top

Table of Contents | Next