In physical modelling using sand, the construction of a sand pack is one of the main experimental difficulties. Existing solutions are generally based either on delicate manipulations (pouring and scraping), or on motorized sedimentation devices, and the qualities of packs thus produced are known to affect experimental outcomes. However, uniformity, planarity, reproducibility and efficiency can be achieved with a simple sedimentation device without motors. A rectangular reservoir pierced with holes rests on a support plate pierced with wider holes. Sand flows when the reservoir is displaced so that its holes match those of the support plate. Sand jets are diffused by planar horizontal sieves, below which the sand sets into the experimental box. Tests on a 250μm median grain size sand show that the density is at its maximum value, reproducible, and uniform. The spatial variations are only of ±0.4% of the mean density. Thickness of the sand layers shows spatial variations around ±2% of the mean thickness, and 6% near the side walls. Very fine grains (90μm median grain size) produce less uniform and less planar packs because of their greater sensitivity to air currents caused by the sedimentation. According to direct shear tests the sand pack has a well defined static friction coefficient decreasing to a lower dynamic value after about 3mm of slip with dilatancy. In contrast, poured sand packs, which are initially less dense, develop only the dynamic friction coefficient (no peak shear stress during slip), without dilatancy. Hole diameters, hole spacing, and the number and openings of the sieves are the parameters controlling the qualities of the sand pack for a given grain size distribution.
Read full abstract