Modeling and propagation of stochastic linear viscoelastic material properties of asphalt mixtures in pavement structures

Pavement structures of asphalt mixtures exhibit variability in their responses to loading and environmental conditions, such as heat and moisture. Thus, it is vital to account for this variability for robust predictions of the response and performance of pavement structures. This study addresses the stochastic modeling and propagation of macroscopic linear viscoelastic constitutive properties of asphalt mixtures within a parametric probabilistic framework. Experimental measurements of the creep compliance have been used to construct two types of stochastic models: spatially-invariant models and spectral stochastic spatially-dependent models of linear viscoelastic material properties. The latter methodology utilizes Karhunen-Loève expansion method for an optimally-reduced stochastic dimension. These models are propagated using a non-intrusive spectral projection method, for which the sparse grid method has been used to sample the cubature nodes. Accordingly, a spectral representation of the response of interest, strain, is constructed and used for statistical quantification. This approach could be appropriate to address various aspects of the performance of asphalt pavements.