| Abstract: | In pavement evaluation using non-destructive testing (NDT), a large amount of deflection bowls are analysed in terms of the elastic moduli of the layers. The results are used to evaluate the material variability, which could serve in an overlay design procedure based on the concept of reliability. The model currently used for interpreting deflection bowls is based on the random variable theory which neglects the spatial distribution of the elastic modulus of the material. Since the subgrade and pavement materials have a spatial distribution, the analysis of NDT could lead to an underestimate of the material variability. The random field theory, which is more adequate than the random variable theory, is presented and used to correct the NDT analysis. The pavement is modelled as a multilayer random elastic medium with, for each layer, a constant Poisson's ratio and a random shear modulus characterized by three statistical moments: average value, standard deviation and autocorrelation function. The stochastic integral formulation presented in the previous publications is generalized here for a multilayer system. The multilayer system is analysed with the random field theory and the covariance matrix of the deflection bowl is obtained and used to generate deflection bowls corresponding to the properties of the random field. These bowls are then interpreted with the current procedure and elastic modulus variabilities are computed. It is found that the current procedure for interpreting deflection bowls underestimates the variability of the subgrade, by a factor of 0.4–1.0. It is interesting to note that the average moduli of the Boussinesq layer and of the two layers are not affected by the type of theory used. The variability of the upper layer in the two-layer system is also unaffected for a small variation range. |
