Learning Forum

[dyy33]

Hello all,

Background

I am currently working on a cylindrical LS-DYNA model that utilize the ALE method and encountering a error output related to negative volumes during advection. The model, shown in Figure 1 (a zoomed in view of the model), is consisted of multiple parts:

1. Blue part is water, modeled by ALE.
2. Red, green and yellow parts are the lagrangian solids. The green part is rigid while others are deformable materials.
3. There is also a void part, overlapping with the lagrangian solids, not shown in Figure 1.

The cylindrical cross section exists in the XY plane and the model is 3 element thick in the Z direction.

[Ian Do]

Hi,nNeg advection vol for ALE elms is caused by a few possible factors, FSI force, BC, prescribed motion, ... If the local force may be too high and pushes more fluid out of an ALE elm in 1 dt than there is fluid available inside that elm to start with, then we have this problem. To fix this we need to find the cause of this local spike of P or force. Then back track to find a way to smooth it out or correct it (in case of FSI force, for ex, we can reduce the coupling P in the load curve). Reducing TSSF sometimes help also.nIf your mesh has very bad aspect ratio as it is here, it may make easier to have neg-adv err in the thin dir (?). n[Q] ... lagrangian solids are more sensitive to meshing quality than ALE?n[A] Not really. All discretized methods depends on the quality of their meshes in general. LAG may be a little more, but I would not be careless with ALE neither.nnI cannot visualize clear the overlapping of the LAG-ALE meshes so I can't comment on the rest. In general, any very thin layers of ALE elms right next to a much larger layer is not recommended. The transition should be gradual. This is a complex model and a lot of details need to be analyzed. This is more of a training issue which cannot be addressed here efficiently (as we cannot trade files). Perhaps you should contact support directly as this can be quite involved. Good luck!n