Per Ullidtz: Dynatest International, Naverland 32, Clostrup, DK 2600, Denmark, Email: pullidtz@dynatest.com
John Harvey: University of California, Davis, California, USA, Email: jtharvey@ucdavis.edu
Imad Basheer: California Department of Transportation, California USA, Email: imad_basheer@dot.ca.gov
Bor-Wen Tsai: University of California, Berkeley, California, USA, Email: bwtsai@berkeley.edu
Carl Monismith: University of California, Berkeley, California, USA, Email: clm@newton.berkeley.edu

| Abstract | Introduction | The WestTrack Experiment | Characterization of Materials | Fatigue Damage of Asphalt |
| Simulation of the WesTrack experiment using CalME | Summary of Analyses and Conclusions |
| Acknowledgement & References |

Characterization of Materials

Some of the most important models used for characterizing the materials are described in the following. Other secondary models are described elsewhere (e.g., Ullidtz et al., 2007). Asphalt Moduli Asphalt moduli were obtained from a number of different test methods. The largest amount of data was from backcalculation of FWD tests done during the experiment. The backcalculation of layer moduli was done using the program Elmod5 (Dynatest, 2005) with a constant non-linearity of -0.2 for the subgrade. Backcalculation of the asphalt layer moduli was done for all of the FWD test series, and for the test positions between the wheel paths as well as in the right wheel path. Indirect tensile tests were done at University of Nevada, Reno (UNR), at “Time Zero Construction” (in September 1995), “Time Zero Traffic”, “12 Months Traffic” and “Post Mortem”, for some of the sections. The moduli were measured at 25 ºC with a 0.1 sec haversine load pulse, but were converted to the reference temperature of 15.4 ºC and 15 msec load pulse duration, corresponding (approximately) to the FWD load in terms of creep test loading time. The observed increase in moduli during the experiment was sometimes very large, in some cases showing a doubling of the modulus. An increase in modulus caused by aging of the binder would be expected to be most pronounced shortly after construction, but there is no increase in modulus from “Time Zero Construction” to “Time Zero Traffic”. This would indicate that most of the hardening is due to decrease in air voids caused by post compaction. Comparison of values at “12 Months Traffic” and “Post Mortem” (i.e., after trafficking had been completed) also indicates that the hardening of the asphalt occurred within the first 12 months of trafficking. Repeated Simple Shear Tests at Constant Height (RSST-CH) were done on the original asphalt and after trafficking (post mortem). These tests confirm the large increase in modulus observed during the experiment. The ratio between the hardened shear modulus (Gpm) and the original shear modulus (Go) is shown in Table 1 for the Fine mix. The hardening is quite similar to what was found from the indirect tensile tests. Table 1: Hardening from shear tests

Frequency sweep tests on beams were carried out by University of California, Berkeley (UCB), but only for one sample of each of the Fine, Coarse and Fine Plus mixes. Shear frequency sweep data were also available for a few of the test sections, at 10 Hz and temperatures of 40, 50 and 60 ºC. Initial moduli were also derived from fatigue tests on beams. Five to six specimens were available for each test section.
Figure 2 compares the asphalt moduli determined by different methods for section 18FHL, which had little hardening and no cracking during the experiment. Equation 14 from Part II Chapter 4 of the NCHRP report (Epps et al., 2002) was also used in the comparison. An asphalt content of 6.2% and an air voids content of 4.3% were used. Different sources give different values for the air voids content but 4.3% appears to be a reasonable value, for section 18, at the start of the test (dropping to about 2.1% towards the end of the experiment).

The legends in Figure 1 are: “Mr HL” – initial resilient modulus from indirect tensile tests (UNR), “FS UCB” flexural frequency sweep data from UCB, “Fatigue HL” – moduli from fatigue beams (UCB), “FS FHWA” – shear frequency sweep data from FHWA, “Table 2.3” – FWD backcalculated moduli from the UCB report (Monismith et al., 2000), “Eq 14-18” – from the NCHRP report (Epps et al., 2000), with asphalt content and air voids for section 18, “FWD” – moduli backcalculated with Elmod5 and “FWD-age” the same moduli adjusted for the effects of hardening. The curve “Model” was the best estimate using the equation:

Equation 1: Master curve format for asphalt.

The format of Equation 1 is the format used in the MEPDG (NCHRP, 2004). Using the MEPDG procedure with the volumetric data for section 18 resulted in a very low minimum modulus (10δ) of 8 MPa and a maximum modulus (10α+δ) of more than 110,000 MPa. Both of these values are unrealistic, and the MEPDG master curve does not compare very well to the measured moduli, as seen in Figure 3.

Figure 2: Asphalt modulus as a function of temperature for section 18, FLH.

Figure 3: Section 18 Best fit “Model” master curve compared to master curve estimated from volumetric data following MEPDG.

An older version of the MEPDG master curve was given by Witczak & Fonseca (1996). This version is also shown in Figure 3 and fits the measured data better than the master curve estimated from volumetric data following the MEPDG procedure.

Unbound Layer Moduli
Triaxial tests were available for the aggregate base and for some of the “engineered fill” lifts, but only for some of the test sections. For the aggregate base, the triaxial modulus was primarily a function of the bulk stress, θ = σ1 + σ2 + σ3, with the shear stress (or deviator stress) having very little effect on the modulus. The modulus could be calculated from:

Equation 2: Modulus of Aggregate Base from triaxial tests.

Where θ is in MPa. The agreement between the modulus measured in triaxial tests and the modulus calculated from Equation 2 is shown in Figure 4.

Figure 4: Moduli calculated from Equation 1 versus moduli from triaxial tests.

Triaxial tests on the engineered fill showed a large variation, with moduli ranging from 20 MPa to 170 MPa.
Analyses of moduli backcalculated from FWD tests showed that the moduli of the unbound layers were also functions of the stiffness of the pavement layers above the layer considered. An example from the first FWD test series on all of the WesTrack test sections is shown in Figure 5.

Figure 5: Correlation between moduli of unbound layers and stiffness ratio of layers above the layer considered (S/35003 in Equation 3), all sections.

The regression equations in Figure 5 result in the relationships:

Equation 3: Influence of confinement on stiffness of unbound layers.

Similar relationships were derived for the aggregate base of the individual test sections for the calibration of the incremental-recursive models of CalME.

The bulk stress was calculated using CalME for section 18 at a depth of 50 mm below the top of the Aggregate Base (AB) and the same depth below the top of the Subgrade (SG, Engineered Fill). Figure 6 shows the calculated bulk stress for AB and SG for the duration of the WesTrack experiment.

 
Figure 6: Bulk stress in Aggregate Base and in Subgrade, 50 mm below the surface of the layers, section 18

Figure 7 shows the modulus of the AB, as calculated from the bulk stress using Equation 2 for triaxial tests and as determined from backcalculation of FWD data. During cold periods where the asphalt layer is stiff, the bulk stress in the AB is low, resulting in a low triaxial modulus. The opposite is true for the FWD moduli.

Figure 7: AB moduli at test section 18, from triaxial tests and FWD.

Figure 8 compares the moduli calculated by CalME, using the stiffness function given in Equation 3, to the moduli backcalculated from FWD testing. Inspection of Figures 6 and 7 reveals that the use of the stiffness function (Equation 3) results in a much better agreement with moduli backcalculated from FWD testing than use of the bulk stress relationship (Equation 2) derived from triaxial testing.

Figure 8: AB moduli calculated by CalME compared to FWD determined moduli.

Figure 9 shows the moduli of different lifts of the Engineering Fill, as determined from triaxial testing at a bulk stress of 30 – 40 kPa, which is on the high side of the actual bulk stress. The moduli from the first FWD tests in March 1996 are shown as a comparison. The average modulus from triaxial testing is 112 MPa, and 81 MPa from FWD tests. Multiplying the FWD derived subgrade modulus by a factor of 0.35, as recommended by the MEPDG (NCHRP, 2004), would clearly not be appropriate.

Figure 9: Modulus of Subgrade from triaxial tests and from FWD backcalculation