Similar to Fig. 8, but for simulation C after 1.0 × 10⁵ years(top) and simulation D after 5.0 × 10⁴ years (bottom). Top panel C): because of the weaker wind the bow shock is very close to the star. The bow shock morphology is completely stable. The shocked gas fills up the cavity behind the star due to low ram pressure of the wind, and thermal pressure of the shocked gas, because the shocks are almost completely adiabatic. Bottom panel D): the bow shock is more conical than for the simulations with a stronger dust component and shows more instabilities. The instabilities are primarily of the Kelvin-Helmholtz type.