Figure 1

Experimental setup. A loaded OS or CS is held on top of a sand bed. It is instantly released.Reuse & Permissions

Figure 2

Penetration depth $z_s$ in function of the load mass $m$ in the four cases: loose, OS (○), loose, CS (●), dense, OS (◇), dense, CS (◆). Values are averaged on 6–10 realizations; error bars show the standard deviation. Data are fitted with Eq. (2). Step plots display quasistatic experiments in dense sand for the OS (dotted line) and the CS (full line).Reuse & Permissions

Figure 3

(a) Scheme of the blown air effect. An air flow fluidizes sand along side surfaces and at the bottom (red circles) of the shell. Volume fraction, penetration, and friction coefficients are locally reduced, which eases the penetration. (b) Snapshots of a glass pushed into a dense sand bed. A movie is available online [21]. The dashed line shows the initial level of the sand free surface. (b1) The initial position. When the glass is pushed into the sand, the level of sand in the glass first decreases a bit as the pressure of the trapped air inside the glass increases. Then, the blown air effect is observed. One sees rising air bubbles at the walls, which eject some sand (b2). Note the string of bubbles; grain ejection is different from the one when an object impacts a granular bed [10, 11, 22]. (b3) The final position of the glass. The level of sand inside the glass has slightly decreased, whereas it has slightly increased at the outer wall. This is not observed for OS (no blown air effect).Reuse & Permissions

Figure 4

Rescaled penetration depth $z/z_s$ in function of rescaled time $t/t_s$. Data include all possible cases with the cylindrical shells, each for three loads (12 experiments) and the steel rod in dense sand (for 3 loads). Symbols are detailed on the figure. All data are well described by a sinus [Eq. (3)].Reuse & Permissions