Résumé
Polygonal Fault Systems (PFS) or networks were generated in 3D finite-difference models within the upper (active) layer of two-layer model. The driving force of this process results from a progressive diagenetically induced volumetric contraction (porosity reduction) of the active layer during its burial. This contraction causes extensional strains, reduces horizontal compressive stresses, and increases deviatoric stresses, leading to elastoplastic yielding and strain-softening of the material in this layer. At a certain point, the initially homogeneous material loses stability resulting in deformation bifurcation and localization within narrow deformation bands (incipient faults). The material undergoes progressive failure within the bands, accompanied by the accumulation of the normal-sense displacement along the faults and the evolution of the fault system, with some faults dying and others forming. The fault architecture characterized in the models by depth-dependent fault density and pattern, is remarkably similar to the natural PFS, both buried and exhumed. The fault spacing (S) in the models scales with the active layer thickness (T) and increases as shear coupling (τ0) between the active and substratum layers decreases. The maximum throw δ along the faults linearly scales with T and increases with the volume or porosity reduction Δϕ. Under the chosen model parameters, the typical δ/T ratio of 0.045 for natural PFS, was obtained in the models for Δϕ ≈ 7%. However, a discrepancy exists between the location (depth) of the δ maximum in the model and nature, which is discussed in the paper and indicates avenues for further modeling development.