[mosse]

Hi

I try to model explosions above a snowpack (for artificial avalanche release). I use a P-alpha eos and an RHT strength and failure constutive model as used for concrete (adapted Drucker-Prager). However, the density (200 kg/m3), stiffness and strength of snow are extremely low. I use an extremely simple axial symmetric geometry (1 meter in x-direction, some meters in y-direction). Nevertheless I encounter many problems with Autodyn:

1) Even with only the material itself (no atmosphere, no loading, no interaction, no gravity), the model does only work with a high energy reference cycle (9999999, "energy error too large" or "time step too small" otherwise) and then still gives stresses although there shouldn't be any. The material even expands in plus x-direction although there is no load.

2) Adding gravity in minus x-direction does not improve things. And the unit of the gravity in Autodyn is not even specified.

3) I need an atmosphere for the explosion. However, if I add the explosion, the pressure (e.g. 100 kPa at sea level) leads to immediate failure within the snowpack. Naturally, snow is porous and would be loaded hydrostatically by an atmosphere which would not lead to failure. I tried to add the atmospheric pressure to snow's strength, but this did not help.

4) If I try to produce a two-layer model (two different snow types), I always get overlapping interfaces. As the geometry is so simple, it is drawn in Autodyn itself.

5) Adding a transmit boundary on the y-end of the model does not help. The material still deforms. The same is valid for a velocity boundary (zero velocity) on the y-end. The velocity is anything else then zero, otherwise the material would not deform on that side and lead to cell degeneration.

6) All different models have one thing in common: Stresses oscillate and there are sharp accelerations, although I haven't used any explosive yet.

Any ideas of how to solve (some of) these issues?

Kind regards,
Stephan