Deflection Response
The agreement between the measured and calculated response during the duration of the WesTrack trafficking on each section, in terms of the deflection under the load plate of the FWD, was very good in most cases. The measured deflections will be a function of the actual asphalt temperature during the FWD tests, the contact between the FWD loading plate and the pavement surface, as well as of the position of the test. The calculated deflection is a function of the following factors which are considered in CalME:

• The estimated asphalt temperature during the FWD test,
• The asphalt modulus versus reduced time relationship,
• The moduli of the unbound materials (aggregate base and subgrade),
• The hardening of the asphalt material as a function of post compaction and ageing, and
• The damage to the asphalt caused by fatigue.

Figure 16: Measured and calculated center deflections on Fine mix sections

Figure 17: Measured and calculated center deflections on Coarse mix sections.

Figure 18: Measured and calculated center deflections on Fine Plus mix sections.

Figure 16 to Figure 18 show the average deflection at the center of the FWD loading plate, calculated for each section and each monitoring session. Only tests done at stations of 30 m or higher were used, in order to avoid the transition sections.
The standard error of estimate is 50 μm for the Fine mix sections, 103 μm for the Coarse mix sections, and 60 μm for the Fine Plus mix sections. For the Fine mix and the Fine Plus mix these values are similar to the standard deviations of the measured values for a single monitoring session on one test section. In other words, the difference between measured and calculated values is similar to the scatter in the measured values. For the Coarse mix the standard error of estimate is somewhat higher.

Relation of Cracking to Damage
Figure 19 compares the damage, ω, predicted by CalME based on the laboratory fatigue data, to the damage estimated from the FWD tests in the right wheel path. As explained earlier, the FWD backcalculated asphalt moduli were corrected for the effects of (estimated) temperature and hardening due to ageing and decrease in air voids content. The difference between the adjusted modulus and the modulus calculated from the modulus versus reduced time model was then assumed to be due to damage.

On average, the Fine, Coarse and Fine Plus mixes all have less damage predicted by CalME than estimated from the FWD. The Coarse mix shows the largest difference, with the FWD estimated damage being 2.2 times that of the CalME predicted damage, even though the shift factor, between laboratory and in situ damage, used for the Coarse mix was 5, compared to 15 for the other two mixes. For the Fine and the Fine Plus mixes the ratio is 1.3 and 2.0, respectively. This indicates that the shift factors are too large, and that they are a function of mix type.

Figure 19: Damage predicted by CalME compared to damage estimated from FWD tests.

An approximate relationship between damage at crack initiation and thickness of the asphalt layers was determined from HVS testing (Ullidtz et al., 2008):

Equation 9: S-shaped curve for damage at crack initiation as a function of AC thickness

Using this equation on the WesTrack experiment would result in a damage of 0.26 at crack initiation. This value appears to be quite reasonable for the Fine mix and the Fine Plus mix. However, for the Coarse mix, the calculated damage was much lower at crack initiation and would correspond better to Equation 10 with the AC thickness raised to -5, rather than to -2, i.e.:

Equation 10: Crack initiation for coarse mix

The cracking (in percent) can be modeled as a function of the calculated damage using an equation of the format:

Equation 11: Cracking in percent as a function of damage.

In Figure 20 the fully drawn curves were calculated from Equation 11 on the assumptions that crack initiation would correspond to 5% cracking and that α was -8.

Figure 20: Cracking models compared to terminal cracking at WesTrack

One of the Fine mix sections, 02FLM, had surprisingly good fatigue performance in the laboratory tests, considering that the AC content was low. The calculated damage is, therefore, low even though the section had a considerable amount of cracking in the left wheel path and some cracking in the right wheel path.

With the large difference between the performance in the right and left wheel paths there is a possibility that the laboratory fatigue specimens were obtained from material that was not totally representative of the section. The calculated damage may also have been underestimated at some of the sections with poor fatigue performance, because the first series of FWD tests were carried out when the sections had already had a traffic load corresponding to 4500 ESALs. Any damage caused by these loads is not included in the calculated damage.

Permanent Deformation
Figure 21 shows the final permanent deformation calculated by CalME (for the right wheel path) and the maximum rutting recorded for the right and left wheel paths, which is not necessarily at the end of the experiment. The average values are: CalME 15.9 mm, 15.8 mm measured at right wheel path, and 18.2 mm measured at left wheel path.

Figure 21: CalME predicted permanent deformation and rutting in right and left wheel paths.

The main conclusions from the simulation of the WesTrack experiment with CalME are:

• The pavement response, in terms of resilient deflections, was predicted quite well, with the difference between measured and calculated values being similar to the scatter of the measured values, in most cases.
• The damage predicted by CalME, using laboratory fatigue tests, was somewhat lower than the damage estimated from the FWD tests. This indicates that the shift factors used in CalME should be reduced. The effect of temperature on damage needs further study.
• The permanent deformation predicted by CalME from RSST-CH testing in the laboratory was close to the measured rut depths.