Hi, I am using the find contacts function with a model made from surfaces. When I do it appears to work but many of the connections are changed to be labeled as “Unresolved_Shell_part…” which also results in errors when running that the set names are too long.
I can’t find the term “unresolved shell” in any documentation
I suspect it has something to do with partitioning.
When the search tool is used to find TIED pairs, it is important to adjust the master and slave items correctly. A single item can only be used once as a master. So the tool tries to resolve this by finding the optimal distribution of master and slave items. If there is no solution to this distribution, an Unresolved list of items is displayed. A new version of PrePoMax has some additional filters to help in such cases.
Using contacts, this should not matter since a single item can be used multiple times as a master item in a contact pair.
Thank you Matej, I suspected it was something like this (I added a bunch of partitions and that error went away)
I ended up having to do the analysis in another program as the model just got too large with surfaces. It would be so nice if you had 1D elements in PrePoMax (with a basic building functionality). Somehow I end up needing them in most of the analysis I do. I have read Calculix beams have some issues but I haven’t spent much time looking into that yet. Maybe I am strange but I end up needing to work with lots of beam type structures (many real world problems) which more or less require 1D beams if they are at all complicated. I love small detailed models but there aren’t that many industries where they can afford the time to do those.
I can only agree with you. Beams will definitely be added. At some point in time. Currently, I have a strong demand for adding Gmsh support for simple hexahedral meshing.
That makes sense, there is something satisfying about a nice clean hex mesh even if modern computing speeds make very fine quadratic tet meshes rather practical and much faster to build. The meshing functionalities in Abaqus are a really nice feature of that program but that is an involved project to replicate.
I noticed this error of unresolved shells too. I did “thicken” a shell mesh and treat it as a solid; but this does not seem to work with other solids being in contact. The Search Contact Pairs shows numerous Unresolved shells. Why would the master / slave distribution matter if I am only after tied contacts?
Sorry, took me a little to break down the issue into a simple (but not pretty) demo file.
Here we have two U profiles. They are drawn as shells. They are bonded to a flat sheet (another shell). It is intended to convert these into layered composite parts later using Calculix keywords.
For manufacturing considerations, the U-profiles would be need to be rounded. In the created area between the U-profiles and the skin layer, an infill is necessary to be placed which has been drawn as solid in this case.
To set-up the model, the shell U profiles have been meshed using “thicken shells”. The meshes part looks like this:
Now, I like to connect (tie) the parts but getting this really long list of ‘unresolved’ constraints. Which appears all the be caused by the solid infill:
Even if accepted, the model will not run. For one, the constraint name is too long; but even if amended the constraints give all sorts of different errors when run as simulation. Haven’t tried to set-up a load case for this demo file but it demonstrates the issue of unresolved constraints:
The file has quickly been sketched using FreeCad 1.0:
Not entirely sure where to go from here :s … but from the above thread, it seems that this is an issue of master/slave surface distribution which cannot be resolved here?!
PS: Just updated the files as initially, there where two equal solids present…
Unresolved constraint means that PrePoMax could not find a solution for master-slave pairs of detected pairs. Calculix allows a surface to be a master surface in any contact only once. It cannot be used in another contact pair as a master surface, only as a slave surface. So PrePoMax tries to find a correct arrangement by swapping master/slave surfaces. In some cases, there is no solution, so the keyword Unresolved is attached to the contact pair name. In such cases, other contact search criteria (try a smaller angle: 1°) or manual contact selection should be used.
Thank you for this.
Initially tired to use the other filters, e.g. only look for solids etc… But changing the angle was indeed helpful! Tried a few different angles and here about 3° seems to actually find the different angles.
The model shown above did actually work, the linked model was modified to look like this and gives all the issues:
Doubt the infill with the very distorted elements is a good idea and will have a good think how to remove these areas because now getting errors like this one:
*ERROR reading *SHELL SECTION: *SHELL SECTION can
only be used for shell or user elements.
Element 404321 is not a shell nor a user element.
or
An element set containing elements with nonpositive jacobian determinant was created.
Many thanks again for the idea of adjusting the angle.
Probably an issue for another topic, only rounding the infills did solve the issue of the model not running even after the constraints were applied correctly. In this example. a 1mm radius was used for above model having 10mm arcs. There are now of course stress concentrations in the areas not supported by material. These areas are smaller than having no infill at all. Not really prepared to create several parts with smaller discretisation in the tips, so I guess I will live with these for now :
As we are living in the world of AI now, here is what Manus.im thought the problem of unresolved contacts was based on the analysis of the issue, the forum and the source code:
These are the UI steps it suggested to try to solve the issue:
And here are other steps which should be considered:
The solution proposed by @Matej was on there, but his suggestion of amending the search angle was only one of many. So, the AI analysis is more of a comprehensive handbook listing all the possible solutions. AI cannot beat human expertise yet Maybe should have asked about the order to try things? Next time…
I’m not sure how long it will be available, but here you can follow the AI reasoning if of interest:
Hope this helps someone else in the future.
I’m afraid I can’t take any credit for the work—this all goes to Diogo: https://www.diogoneves.com.
He has his own YouTube channel where he explores AI applications, and I believe Manus will be featured in one of his upcoming releases. Go and follow him, he may do more Prepomax / Calculix in the future then