General Mechanical

General Mechanical

Topics related to Mechanical Enterprise, Motion, Additive Print and more.

Challenges with Importing Complex Geometries into ANSYS APDL

    • me262
      Subscriber

      Hi,

      I’m trying to import some geometry files into APDL, but it seems impossible to import complex geometries properly. I’ve tried formats like IGESSATParasolid, etc., but none of them seem to work well with complex curved geometries. They work for simpler models, but not for more intricate ones.

      Can anyone confirm if it’s really difficult or even impossible to import complex geometries into APDL? Someone mentioned that it’s possible to create surface geometries in SpaceClaim, import them into APDL, and then convert them into volumes inside APDL.

      If none of that works, I’ll have to create the geometry directly in APDL, but it’s much easier to use something like SpaceClaim for geometry creation—if only the import would work properly.

      How should I proceed?

    • Gary Stofan
      Ansys Employee

      The geometry interfaces into APDL all convert into ANF format. 
      These interfaces and the Ansys Solid Modeling in APDL have not been enhanced in 20 +years. 

      On the other hand, if you are havang issue with all CAD formats, the issue may stem from your original CAD models. 
      Hint: The CAD imports may work better if the original model is upscaled by 10x or 100x. You could then use nscale (downscale) once you are in APDL.

      Discovery, SpaceClaim and DesignModeler can export CAD geometry to .ANF Ansys Neutral file which can import into APDL. 




    • me262
      Subscriber

      I am now able to import the geometry files using both Parasolid and ANF formats, but I occasionally receive the following error:

      *** WARNING *** CP = 122.922 TIME= 20:53:51 Line 47 on area 18 is not on the area within a tolerance. This area could have problems in future Boolean operations. (See the BTOL command).

      Additionally, I am unable to use VGLUE to merge all volumes into a single entity. Visually, the geometry appears correct but is very small (approximately 20mm x 20mm x 20mm) and includes complex curves created with splines.

      I've tried several approaches to resolve this issue, including:

      • Adjusting scaling in ANSYS APDL and creating the geometry in meters to see if working with a larger scale would reduce tolerance issues.
      • Setting the BTOL (Boolean tolerance) to a higher value, aiming to allow ANSYS to accept small gaps or minor misalignments between entities.
      • Re-importing the geometry in different formats (Parasolid and ANF), but the issue persists.
      • Inspecting and refining the geometry in SpaceClaim to check for any gaps or misalignments, followed by re-exporting with tighter tolerance settings.

      Despite these efforts, tolerance issues continue, especially impacting Boolean operations. The main challenge remains the inability to use VGLUE to merge all volumes, which prevents me from progressing with the solution.

      Any additional guidance on further steps or specific adjustments in APDL to enhance Boolean operation success and effectively glue the volumes together would be highly appreciated.

    • Gary Stofan
      Ansys Employee

      Here are some additional suggestions from back in 2004 when APDL Solid Modeling was already in the mature phase. (No further development or defect corrections.)

      What are the known problem areas?
      Some of the known problem areas in solid modeling include:
      - working with “dirty” geometry, either imported or created inside ANSYS
      - merging meshed entities (NUMMRG, xGLUE, xADD, xSBy, etc.)
      - Boolean operations performed on multiple entities at the same time
      - clearing and remeshing (xCLEAR)

      What is the workaround if an error is encountered?
      If a defect or error is encountered, you may CDWRITE the model out and read it back in using the ANF
      format (CDOPT,ANF) to get a “clean” copy of the current geometry. Follow the best practices in the next
      section in order to overcome the initial difficulty or use an alternative method to obtain the same result.

      What are best practices?
      Some of the “best practices” that have been gathered over the years that, in most cases, will yield a
      Boolean-able and meshable solid model include:
      Before exporting the solid model to ANSYS (if applicable):
      -  divide and/or merge the model in the CAD system (rather than in ANSYS) before importing
      -  use tight tolerances in the CAD system before exporting to ANSYS
      - use the CAD system’s geometry checker to verify the CAD data; alternatively, export the
      geometry and re-import it into the CAD system
      -  use the Connection add-on for the CAD system rather than IGES to import the geometry

      In ANSYS:
      -  perform any solid modeling operations prior to any meshing operations
      - use the “input file” approach rather than working totally interactively since input files are
      readily edited and reduce the risk of have only a “corrupted” database to work with
      - if interactive, use SAVE liberally to always have a valid model to go back to in case of
      encountering a failure in an operation
      - SAVE the unmeshed solid model and RESUME it rather than relying on xCLEAR to
      cleanly remove the mesh
      - perform Boolean operations an entity at a time rather than all at once
      - merge nodes before merging keypoints if the entities are already meshed


    • me262
      Subscriber

      Thank you for the quick response.

      Currently, I'm working with Parasolid and ANF files and using vscale. I’ve noticed that vscale might be contributing to some issues. To investigate, I created two simplified test blocks. When I apply vscale to the geometries before using vglue, errors occur. However, if I reverse the order and apply vglue first, everything works smoothly.

      I’ll conduct further tests on my current geometry to confirm if this approach resolves the issue—if I can find the time. If more questions arise, I hope I can continue to receive support here, if possible.

Viewing 4 reply threads
  • You must be logged in to reply to this topic.