ObjectiveEleven concentrated seepage outlets were identified in the dam toe drainage prism. When the reservoir water level reached 180.0 m, two zones of continuous pressurized water boiling emerged on the dam-facing slope of the toe pond.MethodsBased on the geotechnical property parameters of dam materials, a representative dam cross-section was selected for stability simulations via an AutoBank finite element model. Long-term field monitoring datasets, covering downstream seepage outlet positions and measured seepage rates, were adopted to calibrate and optimize the model input parameters. A series of numerical simulations were then performed by integrating various reinforcement countermeasures into the established model, which allowed identification of the optimal risk mitigation and rehabilitation scheme. Practical implementation of this scheme has delivered satisfactory performance and completely eliminated the risk of dam instability.ResultsBefore reinforcement, seepage failure develops at the upstream dam toe, the contact interface between the downstream embankment and drainage prism, the gravel-foundation dam, and the surface clay blanket over the downstream riverbed under the check flood level, normal pool level, and 180.0 m water level, alongside substantial total downstream seepage discharge. After reinforcement, the dam remains seepage-stable under all these hydraulic conditions. For both pre-reinforcement and post-reinforcement configurations, the upstream and downstream slopes meet sliding stability criteria across five hydraulic scenarios: check flood level, normal pool level, 180.0 m water level, rapid drawdown from check flood level to normal pool level, and rapid drawdown from normal pool level to dead water level, without potential sliding or deformation hazards.ConclusionFinite element numerical simulation is adopted as the primary analytical technique to assess the overall stability of dams on gravel foundations. This approach precisely captures the coupled seepage-deformation responses and evolutionary patterns of such dams, yielding accurate, dependable calculation results with well-established, viable technical procedures. Comparative numerical simulations of multiple alternatives enable the optimization of risk mitigation and reinforcement strategies for distressed gravel-foundation dams. This method removes the subjectivity plaguing traditional empirical comparison and delivers a theoretically sound analysis framework. The findings extend the theoretical system describing stability evolution of gravel-foundation dams, and offer robust theoretical support and practical benchmarks for stability mechanism research and refined reinforcement design of similar hydraulic projects globally.