I’m working on this model to obtain a heat transfer and flux diagram but I keep getting this non positive jacobian error. In the last simulation I ran I was able to visualize the result of the failed model and the error is concentrated on few nodes of the geometry. Is there a way to avoid this error or at least to filter out of the simulation these nodes? the forum says I can’t upload the file since I’m a new user, hence I hope a screenshot is enough.
If you want to delete those individual elements from the mesh, you could convert the element set to part and then delete it but it may not be a good idea since those elements may play important role in your model and you probably don’t want to have gaps in it. Unless those are just some artifacts accidentally created during meshing. How was this mesh generated ? Was it imported from other software ? What is the source of the geometry ?
You can use some hosting website like Google Drive, Dropbox or WeTransfer and paste the link here to share the model.
I modeled it on rhino and grasshopper and then exported the geometry on prepomax as an stl file. The mesh was created on prepomax but I can’t see those same spikes in the meshed model before starting the simulation.
I don’t know Rhino but you would probably have to convert this geometry (surface mesh from a 3D scan?) to NURBS. Perhaps this software could also fix the problematic regions that can be causing the negative jacobian error. After exporting from Rhino you can pretty much only try reducing the element size to fit more elements in those narrow regions.
Hi all,
an older topic but I occasionally run into similar issues with non-positive Jacobian elements.
I got a larger model, and when running a static analysis, I get the notification that an element set with non-positive Jacobians was created. I would expect this is a meshing issue and would now change the settings next for the part causing the issue. But the problem I face right now, I cannot locate the elements which cause the error. Is there an easy way of finding these?
For larger element sets, it is straightforward to click on the set and the selection is highlighted in the model. My current model has only 3 of these defective elements.
I simply cannot find these elements in my model being highlighted. So, I have tried to locate the actual elements in the Calculix keyword editor with the intention to get the element if and then use the query tool to locate the issue. But the automatically created data is hidden here and I haven’t found a way to expand the selection.
Is there a way to expand the hidden data in the keyword editor (maybe even manipulate it while we are at it or is there another way of locating the elements belonging to this automatically created set?
Many thanks for your insights!
PS, the question remains but resolved the issue of non-positive Jacobian’s using @FEAnalyst’s suggestions re tied contacts here: Nonpositive jacobian
Just uncheck “Use hiding for faster operation” but you can also export the .inp file and check it manually in any text editor. Or convert the set to a part and show only it.
I had a similar problem in the past, so I thought I would add element nodes to the highlighted items when an element set is selected.
The advantage is that nodes are highlighted by drawing a constant-size circle, which is unaffected by the current zoom. So they are always easy to find. The problem occurs when there are many elements in the element set.
Even after you stated to untick “Use hiding for faster operation” it took me another 5 minutes to locate. It is probably me not finding the obvious staring in my face.
For anyone as lost at me (I might be the only one though), here it is right between the two boxes of the keyword manager:
The existing data cannot be edited though. Guess this is by design?
Thanks @Matej , typically the visual display is clear but in this instance I simple could not locate the elements whatever I tried. I use shell elements and the Calculix keyword manager to create layered materials. Maybe the elements were not yet created? Just speculating but I will keep an eye on it and, if possible, recreate the issue on a simplified model.