By Mike Anderson, P.E. and Dwight Walker, P.E.
In 1996, the Kentucky Transportation Cabinet (KYTC) established a test project to evaluate the in-service performance of various formulations of PG 70-22. KYTC wanted to more completely understand the ramifications of the new Superpave binder system. And KYTC wanted to specifically evaluate if an unmodified binder and binders manufactured using different modifier types and methods performed similarly.
KYTC selected a resurfacing project on I-64 in central Kentucky, about 30 miles east of Lexington, for the test sections. The test area was a rural interstate having relatively constant traffic speed, with no intersections and uniform grades. This section of I-64 is a four-lane, median-divided facility with 33,000,000 ESALs projected over a 20-year design life.
KYTC surveyed its binder suppliers and determined how PG 70-22 binders were produced. The means of producing a PG 70-22 included:
- SBS modification
- Styrene-butadiene-rubber (SBR)/latex modification
- Chemical modification
- Neat asphalt from selected crudes, without modification.
To avoid potential intermixing and contamination with other asphalt materials in the mixing plant storage tanks, KYTC required the contractor to pump the binders directly from the delivery transport to the plant’s asphalt injection system. The test sections were sized to use two transport loads (totaling about 11,000 gallons) of binder. These quantities resulted in approximately one-mile test strips.
Producers for each type of PG 70-22 were identified except for the oxidized material. Two producers of SBS-modified PG 70-22 were identified for a total of five test sections (two SBS, an SBR, a chemically modified asphalt, and an unmodified asphalt.)
As can be expected in field trials, although supplied as a nominal PG 70-22, the properties of the binders, as delivered to the project, varied somewhat. One of the SBS binders actually graded as a PG 76-22.
The mixture used in the test was Kentucky’s standard wearing course mix for moderate and high traffic installations. Although the mix was a Marshall design, it was similar in gradation to a Superpave 9.5 mm (3/8-inch) mix. While this mix did not meet all Superpave mix design requirements, the surface mix had historically shown good performance.
Test Section Comparison
KYTC made an effort to compare the test sections without prejudice. The PG 64-22 control mix was placed in the inside/passing lanes in both directions and in the eastbound direction on a portion of the outside/driving lane. The driving lane control section provided a better performance comparison due to having the same traffic distribution as the test sections. Approximately 22,500 tons of the control mix was used. The five different PG 70-22 sections were placed in the outside/driving lanes.
For this project, the existing 3/4-inch thick layer of open-graded friction course was milled and replaced by a 1.5-inch lift of the surface mix. Before beginning the placement of the PG 70-22 test sections, the contractor was required to produce the control mix for two days to demonstrate that the experimental mix had uniform properties such as gradation, air voids, VMA and density. This uniformity of production was necessary to assure that any performance differences were due to the binder characteristics rather than to varying mixture properties or construction issues.
The contractor began construction on the project in early August 1996 and completed it in mid-September. As part of the acceptance procedure, Marshall specimens were compacted at the plant laboratory for verification of volumetric properties. Lab specimens were compacted using a Superpave Gyratory Compactor for informational purposes. Sieve analyses and asphalt content determinations were performed as part of the contractor’s process control operations.
The variability found was thought to be typical of plant production. The density of the in-place mixes was tested and the roadway cores consistently achieved 93 percent of the maximum theoretical density of the mix produced at the plant. The core density achieved for all the sections averaged 93.7 percent, and the daily averages ranged from 92.1 to 94.9. There were no significant differences in the densities achieved for any of the sections.
A series of site visits was conducted to review project performance. In April 1997, a site visit to the project revealed no rutting, cracking, or other apparent distresses. In 2001, another site visit showed little distress in any of the test sections. This visual observation was confirmed by IRI measurements as shown in Figure 1. For a specific test section, there was little difference in the IRI values from the first measurement (1997) to the last (2004). Also, as shown in the graph, there was little difference in IRI values between test sections.
In 2008, the test sections were generally showing more signs of distress. No noticeable rutting was apparent in any of the test sections. Some sections had a slight amount of raveling between the wheel paths. Most sections exhibited surface longitudinal and transverse cracking. The only exception to this was in Sections 6 and 7 which did not appear to have any significant amount of cracking. Sections 11 and 12 were next best in cracking performance, appearing to have only minor cracks.
Although some sales and marketing people may disagree, it is not important to know which binder was used in which test section. In twelve years, products have changed, different crude sources are used, and different requirements are used. KYTC does not even use PG 70-22 binders anymore, preferring to use PG 76-22 asphalt binder for high traffic loading and PG 64-22 asphalt binder elsewhere.
Did the binder and mix tests conducted in 1996 predict how the pavement sections would perform in 2008?
In 1996, the PG binder specification included a high temperature stiffness parameter, G*/sin d. This parameter was expected to work better than viscosity for modified asphalt binders. Unfortunately, G*/sin d does not correlate well with actual rutting for modified binders. Rutting is a result of damage (high stress and/or strain) imparted by traffic and the response of the asphalt to this damage. The G*/sin d parameter is measured using low stress/strain and does not simulate damage behavior very well for modified asphalt binders.
Recently a new test and specification parameter, the multiple-stress creep-recovery (MSCR) test has been developed, which better simulates the damage behavior of traffic by applying a high stress to an asphalt binder sample and allowing it to recover before starting the process over again. The high temperature specification parameter now being considered by asphalt technologists is the non-recoverable creep compliance (Jnr) of the asphalt binder when tested at the appropriate climatic temperature.
Jnr Rutting Prediction
The data in this table shows the results for three groups of test sections (grouped based on similarity of test values). The mixture rutting results were determined from the Repeated Simple Shear Test at constant height. Although this test is not routinely used now, the same general information can be obtained from a repeated load test using the Simple Performance Test (SPT) equipment.
As can be seen in Table 1, as the Jnr value increases, so does the estimated rutting. From this table, it is important to note two things. First, the Jnr values for all sections are well below the currently recommended specification values. For a PG 70-22 asphalt binder used in a climate that would normally require a PG 64-22 asphalt binder, the maximum allowable Jnr value would be 2 kPa-1. Thus, binder testing would indicate that the asphalt binders used in all sections should be able to adequately contribute to the mixture’s rutting resistance.
Second, the estimated mix rutting from the repeated shear test is low and not practically different for the three groups. This was confirmed by visual observations in 2008.
Thus, the Jnr parameter appears to be a good parameter—for this case study at least—for examining rutting potential. Jnr values were low, leading to the expectation that mix rutting values (tested in the lab and observed in the field) would also be low. As they were.
The PG binder system addressed cracking at intermediate temperatures (fatigue cracking) and low temperatures (thermal cracking). At intermediate temperatures, the specification parameter is G*sin d measured using the dynamic shear rheometer. At low temperatures, the specification parameters are Stiffness and m-value—both generated from the bending beam rheometer.
For the asphalt mixture, the shear frequency sweep test was conducted at intermediate temperature (28°C) to determine mix stiffness. Again, even though shear tests are not often run anymore, the same type of information can be determined using the SPT.
Results for binder and mix testing are shown in Table 2.
Past research indicates that less fatigue cracking could be associated with lower stiffness in the asphalt binder and mixture. The G*sin d values are lowest for Sections 11 and 12 and next lowest for Sections 4, 5, 6, and 7. Mix testing confirms the low stiffness for Sections 11 and 12 as well as Sections 6 and 7, but not Sections 4 and 5.
The mix test data appears to match the visual observations about cracking, but the binder test data appears to be less definitive. It is this concern about the G*sin d parameter that has led the Federal Highway Administration and several researchers to continue to look for a fatigue parameter for asphalt binders.
Finally, it is also somewhat telling that the sections with the least amount of visible cracking (Sections 6, 7, 11, and 12) also had the lowest “true grade” temperature. It is very possible that sometime during the last 12 years of service, the temperature dropped below a critical point such that the pavement sections containing asphalt binders with “true grade” temperatures warmer than -25°C experienced thermal cracking.
Not all of the test sections using PG 70-22 asphalt binders performed exactly the same. Rutting was not noticeably significant in any of the sections. However, cracking was noticeable and significant in most, but not all, of the sections.
Mix testing did a reasonable job of estimating the expected performance in the test sections. Binder testing was less effective, except for low temperature tests, but looks to be more promising with the development of new tests for high (the MSCR test) and intermediate temperatures.
Most importantly, all test sections are still in reasonably decent condition after nearly 12 years of service—regardless of the asphalt binder used.
|Mike Anderson is the Director of Research for the Asphalt Institute. Dwight Walker is the editor of Asphalt Magazine.