INFLUENCE OF STATISTICAL VARIATION IN FALLING WEIGHT DEFLECTOMETERS ON PAVEMENT ANALYSIS

A relatively simple approach that is based on a Monte Carlo simulation procedure is presented to statistically investigate the influence of variation in falling weight deflectometer (FWD) measurements on pavement analysis. The factors considered are pavement moduli and pavement performance or response strains that correlate with pavement performance. From extensive FWD data obtained in Nevada, the variability of FWD deflections was statistically quantified at six sites for two seasons. The FWD tests were carried out at an interval of 50 ft within a uniform 1,000-ft test pavement section at every site. Using these data, as many as 900 FWD sensor deflections with normal distribution for each sensor were randomly generated for each of the 12 cases (six sites, two seasons) studied. The MODULUS program was used in the backcalculation of the pavement layer moduli. The FWD measurements show substantial variation within all of the uniform pavement sections investigated. Backcalculated pavement layer moduli using the generated deflection data show a large variation. The coefficient of variation for layer moduli vary from 5 to as much as 65%. Larger variations were computed for AC and base. However, the coefficient of variations for pavement performance strains are smaller, varying from 8 to 25%. Because the number of loading cycles to cause pavement distress is sensitive to pavement performance strains, the variation in pavement strain caused by the variation in FWD measurements has a significant influence on pavement life predictions.