I have to model a semi-elliptical crack with singular meshing in 3D.
I am able to create half model. But I want a full model with the crack (as my loading is antisymmetric). Hence, I use the reflect command and reflect the volumes with mesh.
After reflecting, I am trying to merge the common nodes in the crack plane, leaving the nodes from the crack region, so as to get a crack.
While merging, I get error- data corrupted, Negative jacobian matrix error. When I plot the error elements, I see the singular elements along the crack front.
Experts I am a novice and not sure if this is the correct procedure.
I am not sure about the issue and partially only understood the issue. Anyway try these...
1. whether the mesh u applied is free or mapped.. If mapped, it won't be uniform everywhere so there may a possibility for not merging the nodes in the crack plane
2. The dimensions of the model is one important factor. If it is very large the part won't get meshed as its elements and nodes increases.
Few other problems are also there in meshing with ansys. Better try in hypermesh with the same procedure and check out the outcome.
Its very easy in ABAQUS, if ur institute has a license for this software
ANSYS is an useful tool in order to develope a 3D full model for cracks under asymetrical loading. Search in Intech open access the lasted Applied Fracture Mechanics Book where you will find a few chapters with novel information.
It all depends on what you are trying to do. If this is a one off analysis where you are only attemptimg to model a single crack configuration then persist with the problem. If you are trying to determine the stress intensity solutions for a family of cracks then using Ansys or Abaqus is not the way to go. (As ana side both use outdated 3D element formulations and you would be better off using NEi-NASTRAN which uses a much more modern element formulation.)
In this case, i.e. where you are seeking to solve for a family of cracks, the whole process is just far too man power intensive and the time taken in redoing the mesh each time far outweighs the compute time. The best way forward to solve 3D fracture mechanics problems is using 3D weight functions. This does not require remeshing for each new geometry. The down side is that you need to have a strong maths background to follow the derivation of the 3D weight functions.
Best of luck.
I solved problems like this before, it works good. Try to solve half the model before if it works good, it means that you have a problem with the merging command.
- Badges
- Science method