|FHWA > Engineering > Pavements > HIF-10-003 > Bulk Specific Gravity of Compacted HMA Specimens|
A Review of Aggregate and Asphalt Mixture Specific Gravity Measurements and Their Impacts on Asphalt Mix Design Properties and Mix Acceptance
Bulk Specific Gravity of Compacted HMA Specimens
Standard Test Methods
The standard test methods for determining Gmb of compacted HMA specimens are AASHTO T 166 and ASTM D 2726. The latter differs from the AASHTO standard principally with regard to its precision statement. Both methods calculate the specific gravity of the sample based on the fundamental density equation, mass over volume. It is therefore important that both dry mass and volume of a specimen be accurately determined. These methods base the determination of the volume of a compacted HMA specimen on Archimedes' principle which equates the buoyant force of an object submerged in water to the volume of water displaced by the object. The problem with this technique is that for specimens with large permeable voids, such as with coarse-graded, gap-graded, or open-graded mixtures, some of the water that enters the permeable voids when the specimen is submerged in water drains out of the specimen when the specimen is removed from the water bath and the surface water dried with a damp towel. The problem is amplified when the air voids in a specimen are interconnected or surface connected, which is often the case with field cores and laboratory performance test specimens compacted to target initial relative densities expected to occur in the field. The result of the water drainage is an error in the SSD mass and thereby the volume determination of the specimen. Consequently, a higher specific gravity value than what the specimen actually has is determined.
Current test methods provide an approach to reducing this error by requiring that specimens with water absorption of above two percent be sealed for testing. For T 166, the method cited for sealing is the paraffin coating method, AASHTO T 275. The ASTM method allows either the parafilm method, D 1188, or the vacuum sealing method, D 6752, when the water absorption exceeds two percent. It is not known why the two percent limit was selected, but is speculated that this limit was determined for Marshall mixes which were typically fine-graded.
Note that the definition of fine-graded and coarse-graded is provided in AASHTO M 323-07, Section 6.1.3.
Precision Estimates of Standard Test Methods
As noted above, the precision information from the AASHTO and ASTM methods are different. The AASHTO method, T 166, only includes repeatability information: "Duplicate specific gravity results by the same operator should not be considered suspect unless they differ more than 0.02."
The precision information for ASTM D 2726 is based on a study conducted by AMRL (2) involving 6-inch (150-mm) laboratory compacted specimens with approximately 4.5 percent air voids. The study included a fine-graded 12.5-mm and a coarse-graded 19.0-mm nominal maximum aggregate size mixture (NMAS) both containing aggregate with less than 1.0% water absorption. The precision estimates from D 2726 are shown in Table 13 and indicate that the method is less repeatable (i.e. higher within-lab precision) for the coarse-graded specimens compared to fine-graded specimens. Potential sources of variation for the SSD method discussed by AMRL include differences in the dampness of the towel used to blot the surface of the specimen, temperature of the water bath, and differences in the interpretations for achieving the SSD condition as quickly as possible.
NCHRP Project 9-26 (11) recently completed a significant study that evaluated the precision estimates for Gmb. The study involved more than 22 laboratories that compacted specimens to 100 gyrations in the Superpave Gyratory Compactor in accordance with AASHTO T 312 then tested the compacted specimens in accordance with AASHTO T 166 and ASTM D 6752 (the vacuum sealing method). Materials variables included two aggregate types (low and high absorption) and two NMAS mixtures. The findings of this study were those mixtures with different NMAS and those containing high and low absorptive aggregate yielded similar precision estimates for Gmb. This study recommended the precision estimates shown in Table 14 for AASHTO T 166.
Alternatives to Standard Test Methods
Several alternative methods available for determining Gmb are listed in Table 15 with their associated advantages and disadvantages.
Several studies have been conducted over the past seven years comparing T 166 to alternate methods for determining Gmb. Many of the studies were sparked by the development of the CoreLok device which is used to vacuum seal compacted specimens in a special plastic bag for a more accurate volume determination when the specimen has interconnected voids.
Buchanan (13) compared AASHTO T 166 with the vacuum sealing method, the parafilm method, and dimensional volume technique. The experimental plan included specimens compacted in the laboratory with an SGC to yield a range of air void contents. Mixture types included coarse- and fine-graded Superpave mixtures, SMA mixtures, and open-graded friction course (OGFC) mixtures. After the Gmb determination was made on the SGC specimens with the four methods, the specimens were saw cut into cube shapes and the Gmb determinations were made again. The study concluded that the vacuum sealing method and AASHTO T 166 provided similar results for fine- and coarse-graded mixtures, but that the two methods gave different results for SMA and OGFC specimens. For these mixes, air void contents with the vacuum sealing method were higher. A good relationship was found between percent water absorbed in the specimens and the air void difference between the two methods. Buchanan also concluded that significant errors can result even when the water absorbed is less than two percent. The final conclusion was that the vacuum sealing method appeared to most accurately measure the Gmb of all specimens regardless of gradation, aggregate type, or compaction level.
Hall, et al. (14) conducted a variability analysis for Gmb determinations using AASHTO T 166, dimensional analysis, and the vacuum sealing method. Field produced Superpave mixtures were collected and compacted in an SGC using between 75 and 129 gyrations according to the mix designs for the field projects. Statistical analyses found significant differences in Gmb results from AASHTO T 166 and the vacuum sealing method. The authors noted that substituting the Gmb results from the vacuum sealing method in place of the results from AASHTO T 166 would increase the calculated air voids from 0.36 to 0.9 percent, and increase VMA from 0.31 to 0.79 percent for the mixtures in the study. Multi-operator variability was also examined. Compared to AASHTO T 166, the vacuum sealing method was found to be less variable for 82 percent of the specimens. Hall et al. concluded that the vacuum sealing method was a viable alternative for determining Gmb. However, agencies were cautioned to consider the shift in Gmb results on calculated mix properties.
Malpass and Khosla (15) evaluated a prototype gamma ray device for determining Gmb and compared the results from this method to those obtained using T 166, the parafilm method, and dimensional analysis. An analysis of variance showed that statistically different Gmb results were obtained among the four methods. It was observed that for mixtures with larger maximum aggregate size and higher air voids contents, the differences between results from the gamma ray device and AASHTO T 166 were greater. Conversely, for specimens with low air voids and smoother surface textures, the Gmb results from these two methods were similar. The authors explained that the Gmb results from AASHTO T 166 were erroneous for coarser, high void specimens due to inaccurate sample volumes caused by the SSD determination. Analysis also showed that the gamma ray method was the least repeatable, followed by AASHTO T 166, parafilm, and dimensional analysis.
Cooley, et al. (16) conducted an interlaboratory study to compare test method precision (single operator and multi-lab) of AASHTO T 166 with the vacuum sealing method. Eighteen laboratories participated in the study. Laboratory molded SGC specimens were made at the National Center for Asphalt Technology (NCAT) and sent to the participating labs. Sample variables were gradation (three levels) and compactive effort (three levels, which yielded essentially three levels of relative density). Results clearly showed that average Gmb results from the two methods were similar for fine-graded specimens, but that AASHTO T 166 yielded significantly higher results for coarse-graded and SMA specimens. The initial analysis showed that a small number of data points were questionable and the investigation found that some problems could be traced back to the specimen fabrication process and discrepancies of sample masses for a few labs. With the explained outliers removed, the statistical analysis indicated that the vacuum sealing method was less precise than AASHTO T166 in most cases. The higher within lab and multilaboratory variability for the vacuum sealing method were attributed to operator inexperience with this method and leaks in the bags (Note that the current vacuum sealing method uses a tougher, better sealing bag). The report discusses at length the precision information provided in AASHTO T 166 and ASTM D 2726 and how they compared with their results. The authors found that their precision results closely matched those from ASTM D 2726 and indicated that the AASHTO precision limits may not be valid. The findings suggest that the vacuum sealing method be used for coarse-graded mixtures when the sample has more than 0.4 percent water absorption. However, for practical purposes, they recommended the vacuum sealing method be used for determining Gmb of all coarse-graded mixtures, including all laboratory molded and field compacted (cored) specimens.
Brown et al. (17) also examined four methods of determining Gmb as part of a larger study. They compared AASHTO T 166, the vacuum sealing method, the gamma ray method, and dimensional analysis. In addition to the four test methods, other experimental variables included field cores, lab molded specimens compacted to three levels of gyration, four gradations, three NMAS, and two aggregate types. Differences among Gmb results with the four methods were found to be statistically significant. Differences between AASHTO T 166 and the vacuum sealing method were small for fine-graded, small NMAS (9.5-mm) mixtures and other mixtures with very low water absorption values. The authors recommended that the water absorption limit for AASHTO T 166 be reduced from two percent to one percent. Although the results suggest that this limit be set even lower, they reasoned that doing so would essentially preclude the use of AASHTO T 166 for most roadway cores. The authors also recommended the vacuum sealing method add a step to reweigh the sample after determining the submerged weight to check for bag leaks. They also advocated a small correction factor of -0.2 percent air voids when using the vacuum sealing method.
Williams (18) evaluated four methods for measuring Gmb, including the T 166, vacuum sealing, dimensional, and gamma ray method using coarse-graded 25.0- and 37.5-mm mixtures compacted to approximately 2, 4, and 7 percent air voids. The results indicated that four methods produced statistically different Gmb results. In addition, T166 had the lowest levels of variability, followed by the vacuum sealing method.
In recently completed NCHRP Project 9-26 (11), a significant part of the study evaluated the precision estimates for Gmb using ASTM D 6752 (the vacuum sealing method). The findings of this study were those mixtures with different NMAS and those containing high and low absorptive aggregate yielded similar precision estimates for Gmb. This study recommended the precision estimates shown in Table 16 for ASTM D 6752. These are greater than the recommended precision estimates for AASHTO T 166 (see Table 14).
In summary, it has been reported in the existing literature that significant differences in measured Gmb using different test methods exist. These differences are more pronounced for coarse-graded HMA mixtures. AASHTO T 166 exhibited the smallest level of variability, followed by the CoreLok, then dimensional method, and finally the gamma ray device.