Verification of Mechanistic-Empirical Pavement Deterioration Models Based on Field Evaluation of In-Service Pavements
This thesis focused on using a detailed structural evaluation of seven (three flexible and four composite) high performance in-service pavements designated as high-priority routes to verify the applicability of the Mechanistic Empirical (M-E) models to high performance pavements in the Commonwealth of Virginia. The structural evaluation included: determination of layer thicknesses (from cores, GPR and historical data), pavement condition assessment based on visual survey, estimation of layer moduli from FWD analysis as well as material characterization. One of the main objectives of this study was to utilize the results from the backcalculated moduli in order to predict the performance of this group of pavement structures using the M-E Design Guide Software. This allowed a quick verification of the performance prediction models used by comparing their outcome with the current condition.
The in-depth structural evaluation of the three flexible and four composite pavements showed that all the sites are structurally sound. The investigation also confirmed that the use of GPR to determine layer thicknesses and the comparison with a minimum number of cores is a helpful tool for pavement structural evaluation. Despite some difficulties performing the backcalculation analysis for complex structures, the obtained results were considered reasonable and were useful in estimating the current structural adequacy of the evaluated structures.
The comparison of the measured distresses with those predicted by the M-E Design Guide software showed poor agreement. In general, the predicted distresses were higher than the distresses actually measured. However, there was not enough evidence to determine whether this is due to errors in the prediction models or software, or because of the use of defaults material properties, specially for the AC layers. It must be noted that although an in-depth field evaluation was performed, only Level 3 data was available for many of the input parameters. The results suggest that significant calibration and validation will be required before implementation of the M-E Design Guide.