Abstract
To address the high computational cost of simulating static stress networks in granular assemblies, this study proposes a compliance-based reduced-order framework with $$\:O\left(N\right)$$ computational scalability. By employing a recursive force-distribution scheme based on local compliance, the method eliminates global stiffness matrix assembly. Structural heterogeneity is governed by a single disorder parameter, $$\:\chi\:$$ , enabling self-consistent force chain formation. The model reproduces fundamental granular phenomena, including exponential contact-stress distributions and the transition between unimodal and bimodal force propagation based on packing disorder. Applied to boundary value problems under rigid indentation, it captures continuum-like reaction stress distributions while accounting for microscopic fluctuations, wall-shear pressure reduction, and localized stress peaks at boundaries. Global energy balance errors remain within approximately 2.5%, and by suppressing widespread tensile forces, the framework effectively captures the macroscopic statistical properties of tension-cut global stiffness matrix solvers without non-linear iterations. This framework offers a low-cost, memory-efficient tool for static pre-failure stress analysis in granular mechanics.

This publication has 23 references indexed in Scilit: