3D coupled dynamic modeling of deep drift excavation near a fault: Coseismic slip and seismic hazards

Seismic hazards induced by deep roadway excavation pose serious threats to underground stability, especially in ultra-close fault environments. To address this, we propose a fully coupled three-dimensional (3D) dynamic modeling framework that resolves excavation-induced stress changes, fault rupture, and seismic wave propagation at near-source resolution. The model quantitatively captures excavation-induced dynamic instability via coseismic slip and seismic moment. Parametric analyses show that fault proximity dramatically amplifies seismic responses: at 1 m fault-roadway distance, coseismic slip reaches 16.7 mm, slip rate peaks at 5.1 m/s, and seismic moment reaches 42.1 × 10 9 N m. Increasing the distance significantly reduces these values, highlighting the role of local stress concentration. Roadway length intensifies dynamic rupture, with the seismic moment increasing to 25.2 × 10 9 N m and the rupture velocity reaching 0.87 km/s for a 60 m span. Roadway rotation exhibits nonlinear effects on fault behavior, with slip decreasing up to 75° and then slightly recovering toward 90°. Strong seismic wave amplification occurs near ultra-close faults, with peak particle acceleration and velocity reaching 120 m/s 2 and 0.44 m/s, respectively. Validation against static solutions confirms the robustness of the model. These results provide quantitative insight into excavation-induced fault instability and guide the design of safer roadway layouts in fault-prone deep mining environments. • A 3D dynamic rupture model coupling excavation, fault slip, and wave radiation was proposed. • The quantities of coseismic slip, seismic moment, and rupture velocity in ultra-close faults were determined. • The nonlinear effects of roadway orientation on fault instability and M 0 were revealed. • Excavation-induced reverse slip under a normal faulting regime was demonstrated. • Seismic wave amplification and roadway hazard were captured via near-source simulation.