Version 9 beta

VictorVictor
edited May 2018
Mecway 9 beta is available at the link below. This doesn't have many major changes but a lot of small improvements. I welcome any feedback on bugs and the changes but probably won't add any new features. For existing customers, it automatically activates if you already have an earlier version activated.

http://mecway.com/download/mecway90beta2.msi
http://mecway.com/download/mecway90rc.msi
Final release of version 9: http://mecway.com/download

Changes, roughly in decreasing order of significance:
  1. 24 bug fixes
  2. Customizable default model allows you to begin with commonly used units/materials/solution variables/etc. (Tools → Options → Default model)
  3. Element axes can be defined by formulas as a function of position. It uses a new interface under Loads & Constraints.
  4. Stress is available for modal vibration without stress stiffening with both CCX and internal solvers.
  5. Faster opening of big files saved with this version.
  6. Radiation in thermal transient for CCX solver (*RADIATE)
  7. Rayleigh damping for CCX solver.
  8. Thermal contact conductance (*GAP CONDUCTANCE) for CCX solver.
  9. Elastic option in bonded contact for CCX. Uses *CONTACT PAIR.
  10. Contact stress and displacement outputs for CCX solver.
  11. Hydrostatic pressure for small displacement with the CCX solver.
  12. Rectangular tube beam section (BOX) for CCX solver.
  13. Line3 beam elements are exported to CCX as B32R instead of B32.
  14. New bending moment solution variable that uses a more consistent sign convention.
  15. Removed Rotate element axes from Rotate/copy because cylindrical coordinates can now be done using a formula.
  16. Tables can be suppressed per configuration so they can be written to different csv files.
  17. Von Mises stress appears by default if no solution field variables are specified.
  18. Button to import gmsh commands from a file.
  19. “Copy data” context menu item for graph.
  20. A new Solution menu that contains all the post-processing items.
  21. Warning if common field variables (displacement, stress, external force, temperature) are omitted with the CCX solver.
  22. Configuration name is appended to the .inp file name for CCX to distinguish multiple configurations.
  23. Paint select brush size is smaller to make finer selecting easier.
  24. Element values are the default instead of node values for stress in element coordinates and beam/truss/spring internal force/moment.
  25. Thermal steady state with radiation and no fixed temperatures automatically uses a lower relaxation factor to help it converge.
  26. Deprecated 2nd orientation for beams in static 2D.
  27. Field variables in the outline tree are listed in the order they were created, not alphabetically.
  28. Solution variables that are only for beams are indicated with “(beams)” on the menu.
  29. Using the hole tool on an isolated node shows a helpful error message.
  30. Mesh tools → Fit to curved surface defaults to cylinder.
  31. Error message for orthotropic and anisotropic material properties that lead to a non-positive-definite material stiffness matrix.
  32. Extrude defaults to 1 subdivision on first use instead of 0.
  33. Spring constant in *SPRING always has a decimal point to suit CCX 2.13’s requirement.
  34. Double click to insert field variable in formula.
  35. Element topology errors (red X) also include collapsed elements.
  36. Beams in deformed view are displayed with their end faces rotated about all 3 axes, not just the longitudinal one.
  37. “longitudinal stress” moved to “stress in element coordinates”
  38. More readable names for *AMPLITUDE, *NSET, etc. in generated .inp files.
  39. Frictionless support in the internal nonlinear solver no longer changes direction to follow the surface direction.
  40. Refine custom shows a warning at or above the maximum number of subdivisions (40).
  41. “Move surfaces together” option in bonded contact rename to “*TIE”.
  42. “tensile force” renamed to “axial force”
  43. Error message for invalid number of nodes in .vol file element.

Comments

  • AndreaAndrea
    Which means "CCX 2.13 features" under LABS options?
    Thanks
  • VictorVictor
    Nothing useful at the moment. Just using *BEAM SECTION instead of *BEAM GENERAL SECTION since the latter seems to be deprecated in 2.13.
  • lk70lk70
    Victor, Appreciate your work to make Mecway perfect.
  • lk70lk70
    the installation package can not be installed on my WIN10 computer, any shortcut ?
  • VictorVictor
    Do you currently have the same problem with version 8 here http://mecway.com/download/mecway80.msi ?

    If not, can you show a screenshot of when it fails?
  • VMHVMH
    Victor,

    It looks like your included ccx version 2.11 is multi-threading although it didn't labeled with "MT". The "CCX Output" showed multiple CPU being used.

    I checked/compared yours against Kwip's complied ccx_2.14_MT on many different linear static analysis and the results are about the same and as fast to solve.

    I also re-checked the linear buckling inconsistent issues we had with "MT" ccx version and it still exist for Kwip's ccx_2.14_MT but doesn't exist for your "ccx.exe". Your are consistent when I run multiple times (more than 10 times with 10 modes). This is great news.
  • VictorVictor
    Thanks for the comparisons. It's not supposed to be compiled with the multithreading option so I don't know what's happening there.

    It's based on the General Electric version. I heard from a former colleague of the guy who compiled it that he's quite smart :)
  • VMHVMH
    Attached are the screenshots of ccx.exe directory and ccx output showing I was using the stock ccx and multiple cpu being used.

    If someone can also verify this, it would be great.
  • VictorVictor
    I think that message might just be reflecting the OMP_NUM_THREADS environment variable. When I change that, it changes the number in the message to match but I still see only 25% CPU usage on my 4-core computer when CCX says "up to 4 cpu(s)".
  • VMHVMH
    I see now. I went back and rerun a large model with about 324k nodes and 8 contacts. ccx.exe took about 5min and ccx2.14_MT took about 3min to solve. Also looked at the CPU usage. ccx2.14_MT used almost at 100% (when using all threads) while ccx.exe didnt.
Sign In or Register to comment.

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

Sign In Register