A Review of Aggregate and Asphalt Mixture Specific Gravity Measurements and Their Impacts on Asphalt Mix Design Properties and Mix Acceptance
Impacts of Specific Gravity Measurements on Mixture Properties
As stated previously, one motivation for adopting a new test method is reducing the variability of test results. An analysis was performed to assess the relative effect of reducing the variability of aggregate specific gravity and compacted HMA test (Gmb) on the VMA of HMA mixtures. The study involved a Monte-Carlo simulation of VMA results calculated using Equation 2; details of the simulation are:
- Values of Gmb and Gsb were randomly drawn from a population exhibiting a normal probability distribution.
- Each simulation included 50,000 calculated values for VMA.
- Baseline "mean" values for the normal distributions were selected to yield a VMA result of approximately 15.1 percent, to represent a typical 12.5-mm NMAS hot-mix asphalt mixture.
- The baseline standard deviation for each of the normal distributions was calculated as the average value of all multi-lab standard deviations (1s) reported by the AASHTO Materials Reference Laboratory (AMRL) for the respective specific gravity. The standard deviation values reported for the traditional SSD method for both Gmb and Gsb were used.
- For Gsb, the specific gravity value used in the simulation was calculated using a 50/50 split between coarse (AASHTO T-85) and fine (AASHTO T-84) aggregate.
- To assess the effect of reducing aggregate and HMA bulk specific gravities on VMA results, the standard deviation of each property was reduced from the baseline value in steps of ten percent, to a final value of fifty percent of the baseline.
Each simulation produced a normal distribution of VMA values. Figure 1 shows the overall result from the simulation analysis. The y-axis represents the variability of VMA, expressed as the standard deviation of the VMA distribution. The x-axis represents the stepwise reduction of the Gmb standard deviation. The discrete points arranged vertically represent the stepwise reduction of Gsb at each x-axis (Gmb) reduction step. Thus, the area bounded by the points shown in the figure illustrates the potential reduction in VMA variability (standard deviation) resulting from reductions in constituent specific gravities.
It is possible to compute the percent-reduction in VMA standard deviation as a function of the reductions in standard deviation of both Gmb and Gsb, as illustrated in Equation 9:
VMAred= 0.4894 (Gmb)- red + 0.4880 (Gsb) red (9)
VMAred = reduction in VMA standard deviation (%),
(Gmb) red = reduction in Gmb standard deviation (%),
(Gsb) red = reduction in Gsb standard deviation (%),
It is apparent from Equation 9 that, in general, the improvement in VMA variability is approximately half (in terms of percent from baseline, or original) that of any improvement in compacted HMA and/or aggregate specific gravity.
The focus on variability (standard deviation) is reasonable in the context of the associated range of two test results. Typically, the acceptable range of two test results is calculated using Equation 10.
d2s = 2.83σ (10)
d2s = acceptable range of two test results
σ = standard deviation of test
In the simulation study, the 'baseline' standard deviation values for Gmb and Gsb yielded a distribution of VMA values with a standard deviation of approximately 1.31 percent. Using Equation 10, the acceptable range of two VMA results would be 3.7 percent. Typical HMA mix design and QA/QC criteria for VMA specifies a total VMA range of only 2.0 or 2.5 percent. Thus, in this example two VMA results which should be considered acceptable could in fact fall outside VMA specifications.
Figure 1 Effect of Reducing Gmb or Gsb Standard Deviation on VMA Standard Deviation.
Changing from T 166 to T 331 (vacuum sealing method) for Gmb determination will also significantly impact several HMA mix properties, including Va, VMA, VFA, %Gmm@Nini, and roadway density, especially for coarse-graded and SMA mixes. Figure 2 shows the relationships between Gmb determined by the two methods from the NCAT study (16). The data are grouped by mix type: fine-graded, coarse-graded, and SMA. The correlation equations between the T 331 and T 166 from this figure are reproduced in Table 17. Using these regression equations, the "corrected air voids" were calculated at two key points in specifications for HMA. According to AASHTO standards, Superpave and SMA mix designs are based on 4.0 percent air voids. Currently, this criterion is based on Gmb determined by T 166. The "corrected air voids" for the three mix types, shown in the third column of Table 17, are the predicted Gmb values if the vacuum sealing method were used. For fine-graded mixes, there is no difference on average, between air voids based on T 166 and T 331. For coarse-graded mixes, the data indicates that when specimens have 4.0 percent air voids based on T 166, the corrected air voids based on the vacuum sealing method would be 4.5 percent on average. Likewise for SMA mixes, specimens calculated to have 4.0 percent air voids based on T 166 would have 4.9 percent air voids when using T 331. Therefore, when using the vacuum sealing method for Gmb determinations during mix design, the air voids and VMA will increase on average by 0.5 percent for most coarse-graded trial blends. This could lead to one of three possible adjustments by mix designers: 1. keep the gradation the same and increasing the asphalt content (~0.2 percent) to reduce the air voids to 4.0%, 2. Increase the dust content to lower air voids and VMA, or 3. Adjust the gradation (shifting finer, toward the maximum density line). Since it may be more desirable to slightly increase asphalt content of these mixes to improve their durability, the first option may be preferred. To assure that this mix design adjustment is selected, agencies may want to consider increasing the mix design VMA criteria by +0.5 percent for coarse-graded mixtures. Similarly, for an SMA mixture, the vacuum sealing method will result in 0.9% higher air voids and VMA on average. To bring the target air voids back down to 4.0%, the asphalt content would have to be increased by about 0.4%. This much additional asphalt could cause problems with rutting and flushing of SMA mixtures. Therefore, it is desirable to balance the change in VMA for SMA mixtures with adjustments in the asphalt content and the aggregate gradation. Therefore, increasing the VMA requirement for SMA by only 0.5% will force a more conservative increase in asphalt content and allow gradations to shift to take up the rest of the VMA difference caused by the vacuum sealing method.
|Mix Type||Regression Equation||Corrected Va for 4.0% air voids based on T 166||Corrected Va for 8.0% air voids based on T 166|
|Fine-Graded||Va(T331) = 0.9884Va(T166)||4.0%||7.9%|
|Coarse-Graded||Va(T331) = 1.1235Va(T166)||4.5%||9.0%|
|SMA||Va(T331) = 1.2312 Va(T166)||4.9%||9.8%|
Using the vacuum sealing method will also significantly change roadway density results for coarse-graded mixtures. Since 92.0 percent of Gmm (8 percent air voids) is a common minimum in-place density requirement in many acceptance specifications for dense-graded mixes, the corrected air void content at this point was also estimated for each mix type. As shown in Table 17, for coarse-graded mixtures, 8.0 percent air voids using T 166 correlates to 9.0 percent air voids (91.0 percent Gmm) using the vacuum sealing method. For SMA mixtures, a minimum in-place density requirement of 92.0 percent of Gmm based on T 166 correlates to a minimum criterion of 90.2 percent if the vacuum sealing method is used. Some agencies require a minimum in-place density of 93.0 percent for SMA mixes to avoid problems with permeability. Adjusting this criterion for the vacuum sealing method yields a minimum value of 91.4 percent.
Figure 2 Comparison of Air Voids for Field Cores Using Gmb determined by AASHTO T 166 and Vacuum Sealing Methods (17).