Granular thermostat implementation within the soft-sphere Discrete Element Method (DEM) framework, considerations and limitations
Dense granular flows are commonly investigated using the soft-sphere Discrete Element Method (DEM), whereas large-scale applications generally require continuum models. Granular temperature, defined as the variance of particle velocity fluctuations, is a relevant variable to collapse rheological scaling. However, experimental forcing approaches used to control temperature also affects other aspects of the granular state. An alternative is to use a thermostat algorithm to control the fluctuation energy directly. Although thermostat algorithms are well established in Molecular Dynamics (MD), their behaviour in dissipative DEM systems has received comparatively little attention. We show that conventional Langevin and Nosé-Hoover thermostats exhibit complementary limitations: Langevin control requires strong coupling to offset collisional dissipation, thereby damping particle dynamics, whereas Nosé-Hoover does not independently disrupt the correlations and segregation generated by repeated inelastic collisions, leading to non-ergodic and numerically unstable states. To address these limitations, we introduce two pairwise hybrid formulations combining deterministic temperature regulation with stochastic decorrelation. Both enforce the prescribed temperature, while variation of the stochastic decorrelation timescale modifies velocity statistics and associated mesoscopic properties such as diffusivity. Finally, application to pressure-controlled simple shear shows that increasing granular temperature at fixed inertial number reduces the apparent friction, consistent with previously reported trends. The framework therefore provides controlled reference states for comparing differently forced granular systems and identifying the variables required for temperature-dependent constitutive models.