Fast-moving accretors are ubiquitous in astrophysics. Their interaction with surrounding gas leaves characteristic imprints, forming structures like bow shocks, Mach cones, and density trails. We study how various physical processes affect the flow structure around an accretor with a one-way surface, its accretion rate, and accretion anisotropy. These processes correspond to distinct length scales: the Bondi radius, the bow shock's stand-off distance, and the Hoyle-Lyttleton radius. We conducted adiabatic hydrodynamic simulations using a spherical coordinate grid centred on the accretor. By varying the accretor's (numerical) size across scales -- from much smaller than the stand-off distance to much larger than the Bondi radius -- we analyse how these spatial scales affect steady-state flow physics. All simulations reach a steady state. When the accretor is smaller than the stand-off distance, a bow shock forms ahead, and a nearly spherically symmetric atmosphere develops within. Accretors smaller than the Hoyle-Lyttleton radius produce a Mach cone, while larger ones exhibit a supersonic-to-subsonic flow transition on larger scales. Fully resolved simulations align with Hoyle-Lyttleton theory, showing slightly anisotropic accretion with enhanced inflow from behind. In contrast, larger accretors approach the geometrical limit, accreting mainly from the flow direction, with a low-density 'shadow' forming behind. The accretor's size strongly influences small- and large-scale morphologies. Resolving the Hoyle-Lyttleton radius is essential for capturing large-scale flow characteristics. Resolving the stand-off distance is needed only to study the bow shock: since it determines the shock's position, its non-resolution does not affect large-scale flow morphology.
We present a visco-elastic coupling model between caked spheres, suitable for DEM simulations, which incorporates the different loading mechanisms (tension, shear, bending, torsion) in a combined manner and allows for a derivation of elastic and failure properties on a common basis. In pull, shear, and torsion failure tests with agglomerates of up to 10000 particles, we compare the failure criterion to different approximative variants of it, with respect to accuracy and computational cost. The failure of the agglomerates, which behave according to elastic parameters derived from the contact elasticity, gives also insight into the relative relevance of the different load modes.