| Popis: |
Estimation of long-term performance (or life) is a critical pavement design concern. Though wheel wander is not routinely addressed in pavement designs, its consideration provides for a more realistic and economical design. The procedures in MEPDG and CalME to address the wheel wander recommend the use of normal distributions for transverse wheel wander. These procedures are based on dividing the wheel wander distribution into a number of segments (say five) of equal areas. Such approaches suffer from a major limitation that the selection of equal segments is arbitrary, can lead to biased results. The study reported here covered a variety of pavement factors that significantly affect pavement performance. These factors included are: (1) pavement layer configuration (thin and thick); (2) pavement material properties (conventional and polymer-modified); (3) tire configurations (dual and wide-base); (4) pavement temperature (T = 70°F and T = 104°F); and (5) vehicle operating conditions (braking and non-braking). A major contribution of this thesis is to provide valuable design information on the relative importance of these factors on the prediction of pavement life. A Monte-Carlo simulation scheme that addressed the role wheel wander on pavement response and performance has been developed. Since the traffic lanes are of limited width (about 12 ft.), the trial values of wheel wander was limited to ± 21 in about the centerline of the traffic lane. The proposed Monte-Carlo scheme provided cumulative distribution functions (CDFs) for all the important responses and they in turn were used in the estimation pavement performance (or life). This study only focused on the impact of wheel wander on HMA failure modes. Required traffic-induced pavement strain database needed in the HMA distress investigations were developed using UNR's 3D-Move model. Three methods were used to evaluate the pavement performance (of life). These methods differ based on the value of the traffic-induced strains used in the performance equations. First method (Method 1) uses the maximum response strain, while the second method (Method 2) is the MEPDG approach. The Method 3 is based on the CDFs developed by the Monte-Carlo simulation scheme described in this thesis. The CDFs were divided into a number of equal segments and the strains that correspond each of the segments were used with the performance equations. As many numbers of segments as needed can be considered, however for being consistent with MEPDG approach (Method 2), it was decided to use five segments. Since the Method 1 uses the largest pavement response, the pavement life predicted by Method 1 is always lower and this may be interpreted as being over conservative. The Method 2 (MEPDG approach), though an important step forwards realistic modeling of long-term pavement performance, its arbitrary use of fixed five locations to define the wheel wander can be biased and therefore, questionable. The Method 3 is statically-based and uses many trials for wheel wander locations to model the vehicle wander. Therefore, it takes into account in a more realistic manner the entire variation in traffic-induced strain on the transverse plane. Such an approach is considered more appealing to pavement engineers and researchers. In summary, pavement design information presented in this thesis are in the form of datasets that the pavement engineers researchers can use to assess the sensitivity of many important factors that affect long-term pavement performance. Neither interpretation nor scrutiny of the design information has been attempted. Instead, the thesis outlines elaborate details on a three-step approach used to develop such design guidelines. |
