Backward erosion piping is a type of internal erosion that has caused the failure of many dams and levees. Despite its significance, reliable prediction of backward erosion piping remains a challenge due to the reliance on average hydraulic gradient methods. In this study, a new, physics-based approach is presented for the prediction of pipe progression. The concept of the critical secant gradient function (CSGF) is introduced as a soil property for predicting pipe progression. A proposed approach for measuring the CSGF in the laboratory is presented, and the measured CSGFs are used in finite element models to simulate four piping experiments with varying geometries, boundary conditions, scales, and soil types. The critical head predicted by the analyses had a maximum error of 25%, with three of the four simulations being within 10%.
