Investigation on jointed rock tunnel stability numerical database construction based on combined FDM–DEM approach

Reliable data sources are essential for intelligent tunnel construction, yet on-site data are often insufficient to meet sample requirements. Previous numerical modeling studies have seldom considered the combined effects of different excavation methods’ spatial effects and the distribution characteristics of joints. This paper develops a method to construct a stability database for jointed rock tunnels with primary support systems using a computational framework combining the finite difference method–discrete element method (FDM–DEM). The framework constructs a 2D model using the Mohr–Coulomb criterion and a 3D model with the Hoek–Brown failure criterion, enabling the stress release process to accurately replicate the influence of joint distribution features and excavation space effects in the 2D calculations by utilizing the longitudinal deformation profile parameters of the bench sections and fine-grained ground reaction curves. The computational circle is determined by grid research and data analysis, while the performance differences of various primary support components and their correlations with surrounding jointed rock are analyzed using the control variable method. The validity of the framework is initially confirmed by case comparisons and macroscopically validated using the Mantel test and Spearman analysis on the constructed simulation database—containing tunnel construction information, joint distribution, rock mechanics parameters, and stability indices—thereby establishing a reliable foundation for machine learning and transfer learning applications.