Linking Microstructure to Cohesive Soil Erosion and Failure

Critical shear stress and erodibility fitted for cohesive soils vary with specimen state and measurement procedure. This structured narrative review examines evidence linking microstructure to erosion and failure, distinguishing cohesive observations, idealized simulations, supporting theory, and transferable analogies. Four testable hypotheses organize the synthesis: H1, prior loading alters the operational erosion threshold through damage and recovery; H2, the location and connectivity of particle loss affect response beyond total mass loss; H3, fabric, chemistry, saturation, and confinement alter detached-unit distributions; and H4, an identifiable evolving-state closure improves independent predictions relative to fixed erosion parameters. The proposed architecture specifies phase-consistent mass exchange, intrinsic permeability, a joint event-to-flux relation, and separate hydraulic-connectivity and mechanical-stability checks. An illustrative pulse-and-recovery calculation connects a lower-tail resistance model to changing thresholds and flux, followed by an independent support-area calculation. These examples specify assumptions and falsifiable comparisons without constituting calibration or validation. Existing evidence supports state sensitivity and distinct detachment modes. Cumulative subcritical cohesive-bond fatigue, universal size scaling, and a percolation-based failure law remain unproven. Independent, multi-observable benchmarks are required to test parameter identifiability, predictive value, scale separation, and transferability.