shock wave through a 2D box of air in Autodyn

  • Last Post 2 weeks ago
zepplin64 posted this 3 weeks ago


In Autodyn 2D, I'm trying to model a shock wave going through a 1000x1000 micrometer box of air (atm pressure). In the center is a circle 500 micrometer diameter representing a bubble of vacuum at about 1 mTorr. I've set up a 2 microsecond square wave shock wave of 1 GPa starting from the left hand side. The mesh is 600x600. The basic expected outcome is to see how a shock wave is dissipated when it encounters a bubble of vacuum in the air. I have the material "Air" set up with the Equation of State (EOS) as "Shock" with appropriate c and s values.

I've tried Lagrange, ALE, and always received errors of a degenerate cell occurring at the boundary of the vacuum bubble although I have tried changing the erosion models. 

Lately with Lagrange, I also receive time step error (being too small), and I have tweaked it to no avail. 

While trying to make a shock wave with a TNT explosion using Euler, I ended up with an overflowing cell (also at the boundary of the bubble).


Any suggestions for possible fixes would be most helpful.



Order By: Standard | Newest | Votes
rwoolhou posted this 3 weeks ago

I'll move this to Mech as Autodyn comes under that product family. As ANSYS staff I suspect we can't comment, but hopefully the community can assist.

SandeepMedikonda posted this 2 weeks ago


Euler stability time step is usually larger than the Lagrange stability time step. So Lagrange Parts impose flow constraints on Euler Parts by covering volumes and faces in the Euler grid. Hence:

  • Lagrange thickness must be larger than Euler cell size for flow constraints to function properly
  • Lagrange element size should be larger than Euler cell size for smoothest coupling
    – 2:1 ratio recommended

Guidelines on the Student Community

zepplin64 posted this 2 weeks ago

Thanks for the response. One of the problems is that I can't set off a predefined shock wave using Euler solver. Do later versions of autodyn allow this? I'm using 2015 Release 16.2 So if I'm using Lagrange, your saying I should try a coarser mesh (fewer elements, thus larger size for each element)?