Try a Mesh > Geometry Preparation
Set both scales to the same size, the mesh density you would like.
Click Ok, and Femap will go through and combine curves and surfaces that would cause problems for the mesh density you set.
Then go back and select the solid(s) and perform a tetmesh, but do NOT change the element density. just click OK.
See if that fixes your bad elements.
Extra time spent preparing the geometry for mesh goes a long ways to getting high quality elements. I have found that while Femap does not give me the best mesh automatically, it gives you a lot of control to get a great mesh.
To find and repair the bad elements I do the following:
To start, I will look only at the Jacobian and check the box for "Make Group with Distorted Elements"
Once that is done, I will add the adjacent elements to the group (Custom Tools-->Grouping-->Elem Group Expand) which makes it easier to see. You can also show the curves so you can see where on the model the problem resides. At that point you can chose to manually adjust the elements, adjust the geometry so Femap has a better chance of producing a good mesh, and/or play with the local mesh controls and refinements/updates.
Before anyone else puts too much effort into this, I helped mesh this for the FEMAP user and was quite happy with getting to a 0.9 Jacobian. I don't want to show the actual model, but it does look something like (b) in this image -
The surface mesh is 662,816 triangles. The CAD geometry was extremely difficiult to deal with, undercuts, voids, all types of problems. I went with the best surface mesh I could, and then cleaned it up to get to the 0.9. I could continue cleaning, however it was my impression that the goal of this analysis was to determine some overall stiffnesses and perhaps some type of crushing analysis, I don't think working the quality of the mesh any further is going to make much difference. The difficulty now is running a a 15M DOF tetrahedral model through a highly nonlinear analysis.
Please don't spend any more effort on how to get a better mesh, it's not a requirement for this model.
Hi All tans and Mashermann,
Thank you very much for the clarifications.
I am totally agreed with you about the problems.and now I am not going to spend more time on it.