New insights into tunnel spalling from scale model and element testing

Brittle instabilities in tunnel excavation can lead to spalling, violent rock ejection, and tunnel collapse. Previous experimental studies have primarily investigated brittle tunnel failure using two-dimensional loading, which oversimplifies in-situ stress state. Additionally, the mechanism behind the low entry angle of fracturing (i.e., angle of spalling fracture at the tunnel wall) observed in spalling during tunneling is not fully understood when using standard uniaxial compression tests (UCT) and triaxial compression (TC) tests. This study aims to showcase these limitations using a large-scale tunnel model and triaxial extension (TE) tests. A tunnel was excavated using a miniature tunnel boring machine (TBM) through an analog brittle rock specimen loaded in a true-triaxial cell. Following excavation, the specimen was loaded in stages under incrementally increasing isotropic loading conditions, which induced failure in the tunnel. The 51-mm diameter tunnel excavated in a 300 × 300 × 300 mm 3 cubic specimen facilitated direct observation of spalling progression throughout the loading stage. Six wideband acoustic emission (AE) sensors were utilized to monitor microcracking intensities associated with changes in boundary stress conditions during the loading stages. At the end of the final loading stage, an epoxy resin was injected into the failed tunnel to preserve the tunnel geometry and identify the damage zone. Triaxial extension (TE) tests were introduced as a more correct experimental procedure to predict tunnel spalling. Using a conventional Hoek Cell, the TE test setup effectively represented the three-dimensional in-situ stress states and yields steep angles failure plane measured from minor principal stress. This study improves our understanding of brittle failure mechanisms based on experimental evidence by comparing the use of TE and TC test results to predict tunnel spalling. The evaluations indicated that predictions using TE parameters were more accurate than those using TC parameters regarding spalling shear strength at the tunnel wall, entry angle, and depth of damage. Direct entry angle measurements from the TE tests and theoretical log-spiral slip lines offered the most accurate fit with tunnel model spalling. The entry angles from the curved failure envelopes from both TC and TE could not provide steep failure plane angles at zero confinement, indicating that a single failure envelope cannot characterize fracturing in rocks at different stress levels. However, the thin shear slabs with steep failure planes provided conclusive experimental evidence confirming that classical shear failure primarily governs spalling. The results provide new insights for safer and more reliable tunnel designs in brittle rocks.