The direct simple shear (DSS) test serves as a vital method in geotechnics, allowing the measurement of peak and post-liquefaction shear strengths, along with the critical state friction angle of soils. Additionally, the simple shearing mode applied in a DSS test is the predominant failure mode in many geotechnical engineering problems. Although the DSS test is widely used to determine soil strength, a significant challenge with the DSS device is the non-uniformity of stress and strain distributions at the specimen boundaries. This non-uniformity depends on not only the specimen size but also the size of soil particles. The influence of specimen size on boundary effects is typically evaluated using the ratio of specimen diameter (D) to height (H). The median particle diameter (D50), as an indicator of a soil's particle size, could be another influential factor affecting the non-uniformities of stress and strain on specimen boundaries in a DSS test. Through three-dimensional discrete element method (DEM) simulations, this research explores these factors. Specimens were generated with a particle size distribution (PSD) scaled from a coarse sand sample. Laboratory monotonic DSS testing results on the coarse sand were employed to calibrate the DEM model and ascertain the modeling parameters. Boundary displacements were regulated to maintain a constant-volume condition which represents undrained shearing behavior. Various specimen diameters were simulated with identical void ratios to investigate the influence of D/H on stress path, peak and post-peak shear strengths, and critical state behavior. DEM simulations allowed the generation of several particle size distributions through different scaling factors applied to the sand gradation to determine the combined effect D50 and D/H. Limiting D/H and D50/D ratios are subsequently proposed to mitigate specimen boundary effects.
