Annotated Literature Review for NCHRP Report 640 (2009)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

228 Table 110: Conclusions from Comparison of LTPP Model Predicted Versus Measured Temperatures Parameter LTPP Model Comments/Explanation High Temperature The LTPP high temperature model at 50 percent reliability provided a close estimation for 2001 within 2.2oC (4ºF). However, in 2002 it underestimated the pavement temperature as much as 5.6ºC (10.1ºF) except for the temperature at 250 mm (10 in.) depth. At 250 mm (10 in.) deep, the LTPP model gave fairly close predictions in both 2001 and 2002. With a reliability of 98%, the LTPP high temperature overestimated the temperature for all cases. Low Temperature The LTPP models at 50 percent reliability did closely compare to measured values in 2001, but overestimated cold pavement temperatures in 2002. In their discussion on the effect of mixes on temperature, Watson et al shows a comparison was made of temperatures at various depths for the OGFC, SMA, and Superpave sections. They show that the layer interface beneath the OGFC (at the middle of the research layers) in July 2001 was 1.7ºC (3.1ºF) cooler than for the SMA and 2.1ºC (3.8ºF) cooler than when Superpave surface mix was used. In 2002, the temperature beneath OGFC and SMA surface mixes was virtually the same at 53.7ºC (128.6ºF) and 53.6ºC (128.5ºF) respectively, while the temperature under the Superpave surface was 55.1ºC (131.2ºF), resulting in a difference of 1.4ºC (2.6ºF). Watson et al mentions that the open surface texture of OGFC and SMA mixes may allow underlying mixes to be slightly cooler than when conventional dense-graded surface mixes are used. The OGFC mix in these sections was placed 18 mm (0.7 in) thick while the SMA and Superpave mixes were placed 38-50 mm (1.5-2 in) thick. 1.58.8 Structural Design No information is provided on structural design. 1.58.9 Limitations No information is provided on limitations of use. 1.59 Watson, D. E., L. A. Cooley, Jr., K. A. Moore, K. Williams. “Laboratory Performance Testing of OGFC Mixtures.” Transportation Research Record No: 1891. Transportation Research Board. National Research Council. Washington, D.C. 2004. 1.59.1 General Watson et al describe the results from a laboratory study conducted with OGFC mixes. The objectives of this study were to evaluate certain criteria which had been previously proposed by NCAT researchers. Specifically, they looked at methods of determination of

229 air voids, design air voids, compaction of mix samples for abrasion loss testing, use of fiber and polymer modified asphalts for reducing draindown, methods of freeze-thaw conditioning for determination of moisture susceptibility and permeability criterion for mix design. Watson et al used three different types of aggregates, three different types asphalt binder, one fiber and Marshall and Superpave gyratory compaction procedures. Watson et al conclude that the CoreLok procedure appeared to be a more accurate method of determining bulk specific gravity of OGFC mixes, the minimum air void content for new-generation OGFC mixtures should be 18 percent based on the dimensional method and 16 percent based on the CoreLok method, the addition of fiber stabilizers significantly reduces the potential for draindown, SGC compacted samples can be used for the Cantabro stone loss test procedure, unconditioned SGC samples should have stone loss of no more than 20 percent, the aging procedure is not necessary for Cantabro test and that one freeze-thaw cycle is sufficient for determination of moisture susceptibility. 1.59.2 Benefits of Permeable Asphalt Mixtures No information is provided on benefits of permeable asphalt mixtures. 1.59.3 Materials and Design Watson et al mentions that three different types of aggregates and three different types of asphalt binder were used in this study. The matrix of materials used is provided in Table 111. Table 111: Materials and Mixes Used by Watson et al Material/property Type Aggregate Granite, siliceous crushed gravel, and traprock Asphalt Binder PG 67-22, PG 76-22 (SBS modified) and PG 76-34 (rubber modified; chemically modified within the refining process). Stabilizer Fiber, 0.4% of the total mix weight. Mix With fiber - PG 67-22, PG 76-22, PG 76-34 Without fiber - PG 67-22, PG 76-22 1.59.4 Construction Practices No information is provided on construction practices 1.59.5 Maintenance Practices No information is provided on maintenance practices 1.59.6 Rehabilitation Practices No information is provided on rehabilitation practices

230 1.59.7 Performance Watson et al provides the results of different tests conducted on the OGFC mixes. These results were used to develop guidelines/criteria for laboratory mix design. The tests were conducted to answer the following questions: 1. Is using dimensional analysis a good procedure to determine air voids in OGFC mixes? A major problem with the dimensional method is that the specimen is assumed to be a smooth-sided cylinder and does not account for the open texture of the sample; specimens have higher calculated voids with the dimensional method than are actually present. 2. Is changing binder grade a better option compared to adding fiber, for reducing draindown? Is the existing 0.3 percent maximum draindown criterion acceptable? 3. Can samples compacted with the Superpave gyratory compactor be used for testing for abrasion loss (according to Cantabro method)? 4. Is the current criterion of 100 m/day appropriate for the mix design of OGFC? 5. Is there a requirement for five day freeze-thaw conditioning for evaluation of moisture susceptibility OGFC mixes during mix design? These test methods are provided in Table 112. Watson et al indicates that there was a very good relationship between air voids obtained from the dimensional and CoreLok methods with an R-square value of 0.84, dimensional method results in air voids were consistently higher than the CoreLok method and that the difference between dimensional and CoreLok air voids increases at an increasing rate as the dimensional air voids increase. From the data the authors infer that if 18 percent air voids is selected as a minimum void content based on the dimensional method, a minimum value based on the CoreLok procedure would be approximately 16 percent. Regarding draindown, Watson et al mentions that the use of fiber stabilizer significantly improved resistance to draindown, much better than increasing binder grade. There was more than 4 percent drain-down in some specimens prepared without fiber while the addition of 0.4 percent fiber by weight of the total mix resulted in minimal draindown. All aggregate-binder combinations met draindown requirements when a fiber stabilizer was added.

231 Table 112: Tests for Different Properties Property Standard Tests and Procedures Used in this Study Air Voids Dimensional volume, CoreLok Draindown NCAT draindown test; draindown was determined for three binder grades – PG 67-22, PG 76-22 and PG 76-34 and three aggregate types. Abrasion/ Stone Loss/ In this study, samples were compacted using both Marshall hammer and the Superpave gyratory compactor (SGC). The SGC samples were compacted to 50 gyrations and the Marshall samples were compacted to 50 blows per face (per standard procedure). Testing was completed using the medium gradation and a PG 67-22 binder and then repeated with a PG 76-22 polymer modified binder. Samples in this study were conditioned by placing them in a forced draft oven at 64°C for seven days. The temperature was set at 64°C to correspond to the average 7-day high pavement temperature for much of the U.S. based on Superpave criteria. The conditioned samples were allowed to cool at room temperature (25°C) a minimum of four hours before testing. Permeability Falling head permeability; method adopted from Florida Department of Transportation Moisture Susceptibility Modified AASHTO T283; Use 30 minutes for vacuum saturation under water and five freeze-thaw cycles for the conditioning process. In this study samples were prepared and tested after 1, 3, and 5 freeze-thaw cycles. The intermediate cycles were added to determine if there was a significant difference in the three conditioning methods. In checking the Cantabro Abrasion loss of OGFC mixes, Watson et al conducted statistical analysis to determine differences (if any) between samples compacted with Marshal hammer and SGC, and between aged and unaged samples (compacted with Marshall hammer). Three replicates were made for each test and results for stone loss were averaged. Watson et al mention that t-test with the Cantabro test results showed no significant difference between aged versus unaged Marshall specimens at a 95 percent confidence level. They infer that if the aging procedure is to be used for SGC samples, a maximum stone loss value should be 24 percent. Again, the t-test for SGC conditioned versus unconditioned did not indicate any significant difference in the data at the 95 percent confidence interval. These tests were run with and without fiber and with and without polymer modifier. The authors concluded that since there are no significant difference between results for aged and unaged samples for both Marshall and SGC samples, the extra week needed for the aging process need not be spent for mix design of OGFC. The authors mention that the use of polymer-modified asphalt made a significant difference in results for the Cantabro Abrasion loss test. Abrasion loss was reduced considerably as the PG binder grade increased in stiffness for both conditioned and unconditioned samples. Tests with the PG 76-34 binder had virtually no stone loss for both conditioned and unconditioned samples. Regarding permeability test results, Watson et al mentions that three replicates were made and results were averaged for each reported test. Sample combinations that met draindown and Cantabro requirements were tested for permeability using a falling head permeameter.

232 Permeability tests were performed using the procedure adopted by the Florida Department of Transportation for mix designs with specimens compacted at 50 gyrations with the SGC. They mention that permeability values failed to meet the criterion of 100m per day (minimum) in several cases. Based on previous research, permeability values of at least 100 meters per day were recommended for acceptable performance. Failing results were typical especially for the fine-graded blends with the exception of the traprock mixtures. Generally, the permeability can be increased by making the aggregate gradation coarser. However, when the coarser gradation was used in this study, it also resulted in a greater potential for draindown problems. Watson et al indicate that for the fine gradations used in this study the 100 meters per day criterion may be difficult to achieve for some aggregate types. They also mention that the permeability test, however, was highly variable and had a within lab standard deviation of 22.76 m/day. Also, draindown of the asphalt binder through the specimen may account for some of this variability, since draindown caused some of the samples to become sealed around the bottom of the specimen so that they were impermeable. For specimens where this was discovered, samples were remade. The authors also mention that aggregate type was also found to have an effect on permeability results. They explain this with the fact that the total volume of aggregate in the sample will change with a change in aggregate bulk specific gravity. From the data the authors infer that dimensional air voids would have to be about 20 percent in order to meet the standard permeability of 100 meters/day. The 20 percent air voids is consistent with values being used in Europe. In order to meet permeability of 100 meters/day, the CoreLok air voids would have to be approximately 18 percent. Watson et al mention that in evaluation of moisture susceptibility, very little overall change was observed (no significant difference was observed in Analysis of Variance, ANOVA) between the three conditioning methods used (1,3 and 5 day freeze-thaw) for samples with PG 76-22 binder and fiber stabilizer. By using only one freeze-thaw cycle for the moisture conditioning process the time needed to perform an OGFC mix design can be reduced by about two weeks over the previous method that recommended five freeze-thaw cycles. They show that both the granite and gravel mixtures exhibited a slight tendency to gain strength with additional freeze-thaw cycles although the traprock showed a slight decrease in strength. These samples averaged more than 100 percent retained tensile strength when compared to unconditioned samples. Watson et al showed that tensile strength values for mixture using PG 76-34 binder and fiber stabilizer were significantly lower than the samples with PG 76-22 binder, and that, the tensile strength was virtually unchanged as the number of freeze-thaw cycles increased. The tensile strength ratio of conditioned to unconditioned samples was higher than the minimum recommended value of 80 percent in each case. Based on the assumption that 80 percent retained strength is a criterion for identifying good and poor