Path dependency in multistage direct shear tests on a dense mixed-grained soil: a DEM-based micromechanical interpretation

Multistage direct shear testing reduces material demand and specimen-to-specimen variability compared with conventional testing, but later stages may become path-dependent because the specimen is not fully reset between loading steps. This study investigates this path dependency for a dense mixed-grained gravel–sand with plastic fines (GKB1) by combining laboratory tests with a DEM-based micromechanical interpretation. A dataset comprising 24 single-stage (SS) and 35 multistage (MS) tests was analysed for different multistage procedures (MSA–MSE) at densities around \(D_{\textrm{Pr}}\approx 98\%\) . Clear method-dependent deviations emerge in later stages, particularly at \(\sigma'_N = 400\,\textrm{kPa}\) . These deviations affect shear stiffness and maximum dilatancy more strongly than peak strength. While the initial consolidation interval (0–100 kPa) shows highly reproducible behaviour, the final interval (300–400 kPa) reflects accumulated history effects prior to the last shear stage. A three-dimensional DEM model, calibrated against the SS response, was used to interpret the internal origin of these deviations within a simplified coarse-grained load-bearing skeleton. The simulations indicate that successive loading stages progressively alter fabric, contact organisation and force transmission, generating inherited internal states that influence later-stage response. Overall, the combined experimental and numerical results show that multistage direct shear can reproduce peak-strength trends reasonably well, but stiffness, dilatancy and internal structure remain strongly path-dependent. Multistage direct shear tests reveal loading-history effects on stiffness and dilatancy, while DEM simulations identify method-dependent changes in contact fabric