This study involved the determination of the structural layer properties of aggregate and asphalt surfaced pavements for use in the evaluation of pavement load carrying capacity and the seasonal effects of moisture. This was achieved by the use of Nondestructive Testing Methods (NDT) and backcalculation techniques. Pavement surface deflection measurements were taken using three NDT devices, namely the Dynaflect, Road Raterand Falling Weight Deflectometer (FWD). Soil moisture cells were implanted in the subgrade to monitor the moisture content and temperature of the subgrade. Samples of the pavement materials and subgrade were taken for evaluation in the laboratory using standard methods. A total of 27 project sites were studied. The deflection data collected using NDT devices were used to backcalculate pavement layers and subgrade moduli. The backcalculation procedures used were BISDEF, MODCOMP2, and SEARCH. The backcalculated moduli obtained from different equipment for any test site were compared to each other. Also, the backcalculated moduli were compared to laboratory determined values. Attempts were made to develop regression equations relating subgrade moisture content to easy-to-measure variables, such as rainfall, site aspect, pavement thickness and elevation. Also, for a known stress state, attempts were made to develop a regression equation which can be used to predict modulus as a function of dry density and moisture content. Finally, a regression equation was developed for the prediction of backcalculated modulus from pavement surface deflection basin for aggregate surfaced roads. The results from the study show that backcalculation methods can be used to determine the moduli of aggregate and asphalt surfaced pavement layers. These methods are limited in that they cannot take into account the stress sensitivity of the pavement materials and subgrade. As such, the backcalculated modulus denotes the relative stiffness of the pavement layer and cannot be compared to the modulus obtained in the laboratory at a specific loading situation. Nevertheless, it was shown that the backcalculated modulus changes with the moisture content and, therefore, can be used to evaluate the seasonal effects of moisture. The results can be used in the management of low volume roads, first by providing a means of evaluating the in situ pavement conditions at any given time. This can be used to develop rational load restriction policies during the critical season of the year. The backcalculated moduli can also provide a reliable input in overlay and reconstruction design. Finally, the prediction equations can provide an easy method for obtaining the resilient modulus for input into a mechanistic design method.

As with the previous two symposia, the 32 papers from the June/July, 1999, Seattle symposium present advances in the nondestructive testing of pavements using conventional falling weight deflectometer techniques and other promising techniques such as ground penetrating radar, rolling weight deflecto

None of the existing backcalculation procedures is guaranteed to give a reasonable, effective elastic layer modulus for every deflection basin measured. The differences of results among various backcalculation programs are magnified by the differences of distinct analysts using the same program but applying different input parameters. A more experienced analyst often may obtain better results. Some of the difficulties, such as the deviation of pavement material properties from the linear elastic, isotropic, and homogeneous assumptions, the uncertainty of input parameters (for example, the deflection measurement), and the limitation of basin matching techniques, may only be dealt with by informed engineering judgement. A knowledge-based expert system approach as described in this paper may be used to generate more consistent and better estimations of in situ pavement material properties by combining numerical computations with a pavement expert's knowledge and judgement. A knowledge base which serves as an intelligent pre- and postprocessor to the backcalculation program is developed. The knowledge base may be revised easily as better understanding emerges.