The multi-material ALE formulation is used to model the response of a plate subjected to close-range explosions in LS-DYNA. Both the explosive and the surrounding air was explicitly modelled.The coupling between the fluid and the structure was established based on a penalty coupling method. However,an obvious leakage of both the pressure and the explosion products was observed during my simulations. The penalty factor PFAC was increased to attempt to improve leakage. When the leakage is improved a little bit, however, another problem arises. The sliding energy reaches a very large negative value compared with the internal energy, and it means the result is incorrect. Do these problems need to be taken care of when simulating explosions? What should I do to get an accurate result through a numerical simulation in LS-DYNA?
A few things to check:
- Are your shell or segment normals facing the fluid. This must be done.
- Mesh discrepancy between your Lagrangian and Eulerian parts should not be too large ideally 1:1 ratio.
- Create an *ALE_MULTI-MATERIAL containing all the AMMG's of interest and couple to that.
- Avoid mass scaling as the advection errors quickly increase with large time-steps.
- Correct EOS for all fluid parts (i.e. air and explosive) are a must
A few things to check:
- Are your shell or segment normals facing the fluid. This must be done.
- Mesh discrepancy between your Lagrangian and Eulerian parts should not be too large ideally 1:1 ratio.
- Create an *ALE_MULTI-MATERIAL containing all the AMMG's of interest and couple to that.
- Avoid mass scaling as the advection errors quickly increase with large time-steps.
- Correct EOS for all fluid parts (i.e. air and explosive) are a must
Dear Michael Saleh,
Thank you for your reply.
I think I have done simulation as you said. However,the leakage problem still exists. Is there any other comments?
Your Sincerely
XinFengDuan
- Manually define your coupling pressure, PFAC, using the penetration depth as your abscissa and the pressure as your ordinate. (can severely slow you simulation)
- increase the NQUAD number (slows your simulation)
- use ILEAK option
- In *control_ALE vary the advection method but keep the NADV value to 1
- Badges
- Science method