U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations
|This report is an archived publication and may contain dated technical, contact, and link information|
Publication Number: FHWA-HRT-05-079
Date: May 2006
Optimization of Traffic Data Collection for Specific Pavement Design Applications
Chapter 5. Sensitivity Analysis
Effect Of Traffic Data Collection Scenario
The 30 LTPP sections described in tables 11 and 12 were used in conducting the detailed sensitivity analysis of the NCHRP 1-37A design guide related to traffic input. The traffic data were simulated according to the scenarios described above. The NCHRP 1-37A design guide analysis was conducted using site-specific layer thicknesses and climatic data as summarized in Appendix C; however, no site-specific pavement layer moduli were used because the available data were in the process of being reevaluated at the time of this study. A total of 1,950 NCHRP 1-37A design guide runs were planned (i.e., 30 pavement sites x (5 continuous-coverage scenarios + 12 variable-coverage scenarios x 5 reliability levels)). For the variable-coverage scenarios, five runs accommodate only one-sided reliability (i.e., traffic underprediction, which is critical), as explained earlier. However, early in the sensitivity analysis, it was concluded that only scenario 1-0 is capable of providing 99.9 percent reliability. As a result, NCHRP 1-37A design guide simulations for this level of confidence were not conducted for all of the sections, nor were they considered in developing the final traffic data collection recommendations. Instead, a number of NCHRP 1-37A design guide runs were conducted to simulate the performance of some of the selected sections on the other end of the confidence interval (traffic overprediction) to establish the complete range of pavement life prediction estimates. NCHRP 1-37A design guide pavement-performance predictions are in terms of particular distress parameters versus time. In defining pavement life, the limiting values of these parameters had to be assumed, as described in table 24.
Pavement life was defined as the length of time it takes to reach the limiting value for one of these distresses, called the "critical distress parameter." An example of these pavement life predictions is shown in figure 9 for section 181028, which was estimated to fail in rutting. Note that for clarity, only the results from the continuous-coverage scenarios are shown. For the particular site, it can be seen that the traffic data collection scenarios do have a significant effect on estimated pavement life. For this example, pavement life ranges from 26 years for scenario 3-0 to 43 years for scenario 4-4, but where the true estimate (i.e., obtained under scenario 1-0) is 28 years.
Figure 9. Example of NCHRP 1-37A design guide output, site 181028 in Indiana.
Obtaining pavement life from the output of the multitude of NCHRP 1-37A design guide simulations performed was automated using a macro that identified the critical pavement distress parameter and the number of years that it took to reach it. A summary of the estimated lives in years under scenario 1-0 and the critical distress parameters are given in tables 25 and 26 for the flexible and rigid sections, respectively. To facilitate interpretation of the pavement life prediction results across scenarios and confidence levels, it was decided to focus on a particular distress parameter, selected for each site, to be the one critical parameter under scenario 1-0. Furthermore, this approach allowed testing of the sensitivity of individual damage models to traffic input. As mentioned earlier, the life predictions in these two tables were obtained under an assumed annual AADTT growth rate of 4 percent. For each of the 30 sites analyzed, the results were summarized by plotting pavement life versus traffic data collection scenario by confidence level (figures 10 through 14).
Figure 10. Summary of mean in life predictions, site 181028 in Indiana, confidence 50 percent.
Figure 11. Summary of the range in predictions, site 181028 in Indiana, confidence 75 percent.
Figure 12. Summary of the range in predictions, site 181028 in Indiana, confidence 85 percent.
Figure 13. Summary of the range in predictions, site 181028 in Indiana, confidence 95 percent.
Figure 14. Summary of the range in predictions, site 181028 in Indiana, confidence 99.9 percent.
These plots show life predictions for both ends of the confidence interval to offer an idea of the range in pavement life predictions obtained from the discontinuous data coverage scenarios. Appendix D contains a series of tables summarizing the estimated lives by section for all scenarios and confidence levels.
Effect Of AADTT Growth Rate
The sensitivity analysis conducted so far considered that the annual growth rate in AADTT was 4 percent for all of the scenarios simulated. Additional analyses were conducted to establish the actual annual growth rate in AADTT for the sections that had multiple traffic data years. A compound traffic formula was used for this purpose. Pavement life predictions were obtained with the NCHRP 1-37A design guide software using the actual AADTT growth rate under scenario 1-0 input. The actual growth rates calculated and the resulting pavement life predictions are shown in tables 27 and 28 for the flexible and rigid sections, respectively. In general, the actual annual AADTT growth rates differed significantly from the assumed value of 4 percent, ranging from -29 percent to +28 percent. Where actual annual AADTT growth rates were estimated to be negative, they were assumed to be equal to zero in predicting pavement performance. Furthermore, it was not possible to capacity constrain future AADTT, where the calculated annual AADTT growth rates were unusually high (i.e., the NCHRP 1-37A design guide software does not allow changing of the traffic growth rates during the analysis period). A comparison of the resulting pavement life predictions between scenario 1-0 with the assumed 4 percent annual AADTT growth rate and the actual annual AADTT growth rate are shown in figures 15 and 16, respectively.
Figure 15. Pavement life prediction comparison between actual annual AADTT growth rate and 4 percent annual AADTT growth rate, flexible pavement sites.
Figure 16. Pavement life prediction comparison between actual annual AADTT growth rate and 4 percent annual AADTT growth rate, rigid pavement sites.
As anticipated, the greater the difference between the annual AADTT growth rate and the assumed 4 percent value, the greater the difference in estimated pavement lives. These differences in pavement life were as high as 256 percent (i.e., for flexible section 181028).