Spectral contact networks for granular base load spreading

Unbound granular bases redistribute traffic loads through heterogeneous particle contacts, but pavement-scale analyses rarely provide a lightweight operator linking contact-network structure to transmitted base-reaction footprints. This study proposes a spectral contact-network surrogate for moving-load transfer in granular base layers. A compacted assembly is mapped to a stiffness-weighted contact graph, and an energy-based reduction under vertical kinematics yields a sparse grounded Laplacian system coupling internal contact transmission with subgrade support. The operator directly outputs reaction footprints and spreading metrics while retaining interpretation in terms of contact topology, stiffness heterogeneity, and spectral bottlenecks. Moving loads are represented by repeated quasi-static solves along the top boundary to generate footprint maps, peak reactions, spreading widths, localization metrics, and entropy-based dispersion indicators. Spectral diagnostics identify weak partitions and connectivity-critical edges, while proper orthogonal decomposition (POD) shows that the footprint evolution is strongly low-rank, with approximately five modes capturing 90% of the footprint energy and six modes capturing 95%. Comparison with discrete element method (DEM) simulations for localized compaction shows that the framework captures the dominant directional kinematic signature. Computationally, each DEM case required about 20 min, whereas the proposed framework completed the corresponding case in less than 10 s after assembling the contact graph and grounded system. The results show that load spreading is governed jointly by contact-network redundancy, support constraint, and load position, and that targeted removal of connectivity-critical edges causes stronger deterioration than random damage. The framework provides an efficient and interpretable intermediate-scale tool for granular load-transfer analysis beyond homogenized continuum descriptors.