The results from WesTrack were imported to the CalME database and the experiment was simulated, section by section, using a time increment of one hour, and the measured temperatures and load applications during each hour. It is very important that the pavement primary responses (stresses, strains and displacements) are predicted reasonably well for the duration of the experiment. If the responses are not correctly predicted it will not be possible to calibrate the empirical relationships.

To predict the pavement responses for the duration of the experiment, it is necessary to consider any fatigue damage that develops during the testing, as this will influence those responses. Getting a good match between measured and calculated responses, therefore, implies an adequate calibration of the fatigue damage functions.

Pavement Response
The only measured response available from WesTrack was the FWD deflections. Figure 10 shows the FWD deflection at the center of the loading plate in the wheel path (position F3) obtained for Section 18 at various times during the WesTrack experiment. The deflections correspond to a peak load of approximately 40 kN (the actual load level was used both for measured and calculated values). The legends marked “M” are the measured values. The measured points are connected by fully drawn lines. The corresponding calculated deflections have legend “C” and the points are connected by dotted lines. Figure 10 shows results from four FWD positions. “35_1” in the header of the figure indicates a test point between station 30 m and 39 m.

Figure 10: FWD deflections at section 18 (in wheel path, geophone under the loading plate)

The agreement between the measured and the calculated deflections is seen to be very good in this case. The mean difference between the four measured deflections in Figure 10 is 2 μm (10-6 m) and the Root Mean Square (RMS) difference is 55 μm. The mean difference between measured and calculated deflections is 3 μm and the RMS is 33 μm, so the scatter in the measured deflections is as large as the difference between the measured and calculated values.

Figure 11: FWD deflections at section 18 (between wheel paths, geophone center of load plate)

Figure 11 shows the deflections measured between the wheel paths, compared with the calculated deflections based on the damaged asphalt in the wheel path (calculated deflections are identical to the calculated deflections of Figure 10). The figure clearly shows that the asphalt in the wheel paths did suffer some damage, resulting in larger deflections, even though no visible fatigue cracking was recorded on this section.

Figure 10 shows that the deflections calculated by CalME, using the master curve for the asphalt material and the models for determining the stiffness of the unbound materials, and considering the effects of fatigue damage, are in good agreement with the measured deflections. This is a good indication that the response model is functioning correctly, and that the predicted stresses and strains will also be reasonably correct, so that they can be used for calibrating the empirical models dealt with in the following sections.

Fatigue Damage
Section 18 had no visible cracking during the experiment, but still suffered some fatigue damage, as indicated by both the FWD backcalculated moduli and the simulated damage (parameter ω in Equation 5) shown (on the left axis) in Figure 12. Both are from the right wheel path (there were no FWD tests done in the left wheel path). The damage from FWD backcalculated moduli was calculated on the assumption that any difference between the backcalculated modulus and the modulus from the master curve, adjusted for temperature and hardening, was due to damage. The cracking in the two wheel paths is shown at the axis to the right (in this case there was no cracking, as mentioned above).

Figure 12: Damage in right wheel path of Section 18 (LWP left, RWP right wheel path)

Figure 13: Damage in right wheel path of Section 16

The damage predicted for Section 16, which had low binder content and high air voids, is shown in Figure 13. This section had a few percent cracking in the left wheel path (LWP) and about 50% in the right wheel path (RWP), at the end of the experiment.

Permanent Deformation
Figure 14 shows the down rut in the right wheel path of Section 18. Again, measured values are connected by a fully drawn line whereas the values calculated by CalME are connected by a dotted line.

Figure 14: Down rut in right wheel path at Section 18.

In Figure 14 the mean difference between the measured down rut depth and the calculated permanent deformation is 0.9 mm and the RMS is 1.3 mm. Figure 15 shows the simulation results compared to the measured maximum rut depths (the distance from the highest peak of the rut to the bottom of the rut) in the right and left wheel paths. The mean difference between rutting in the left and the right wheel path in Figure 15 is 2.7 mm and the RMS is 4.0 mm.

The maximum rut depth in the right wheel path, shown in Figure 15, is not much different from the down rut shown in Figure 14, whereas the rutting in the left wheel path is larger.

Figure 15: Maximum rutting in left and right wheel paths at Section 18.