Forensic back-analysis of a reactivated deep-seated landslide in volcanic residual soils of the Cisumdawu highway KM 178, Indonesia

This is the authors' abstract. We don't add key points for this paper.

Background This study is a forensic back-analysis of the 2021 embankment failure at KM 178 of the Cisumdawu Highway, Indonesia. The post-failure investigation identified weak subsurface materials but no discrete failure horizon, and the mechanism was never established. Six independent lines of field evidence were assembled, from UAV LiDAR geomorphology to paired triaxial and direct shear testing. A two-dimensional Morgenstern–Price model was built with the failure surface fixed from that evidence while material parameters were calibrated. Pore pressure was represented as a material-specific ratio rather than a piezometric surface, the residual soil permeability being too low to transmit storm rainfall to depth. Uncertainty was propagated by Monte Carlo simulation. Results The alignment crosses a valley floored by relict landslide deposits, with arcuate paleo-scarps on the upper slope. The weathering profile grades from residual silty clay through residual silty sand to weathered tuff, placing a granular horizon within low-permeability clay, so permeability is not monotonic with depth. Inclinometers installed after the rupture intercepted a shear horizon at 559 and 536 m elevation, recording 15 cm and 10 cm of cumulative wet-season displacement, while the instrument monitored during the dry season recorded none. Deeper horizons remain active but could not be traced laterally, indicating a larger relict complex. That horizon coincides with N-SPT below 10 enclosed within material returning 20 to 30, a strength inversion weathering cannot produce, and with gravelly deposits of low resistivity. Without imposed pore pressure it is not the critical surface, returning 1.40 against 1.29 for a shallower surface within the embankment. It becomes critical only when the pore-pressure ratio reaches about 0.37, at an operational friction angle of 20.5°, below the measured peak of 29°. Sensitivity analysis identifies that ratio, not embankment strength, as the dominant control. Conclusions The failure is best explained as rainfall-driven reactivation of a relict shear surface, recharge reaching it laterally through the granular horizon rather than by vertical infiltration. The relict surface was the necessary condition and the pore-pressure rise the sufficient one. On relict landslide terrain, mitigation should target drainage of the basal horizon rather than strengthening of the embankment.