Version 24

Victor

Mecway 24 is now available at the link below. As usual, it's a free update for all existing license holders, and includes the Spanish language manual translated by Germán Bresciano.

https://mecway.com/download

Changes

• Symmetry boundary condition.
  
• Node-surface coupling works with the internal solver and includes distributing (RBE3).
  
• CCX updated to 2.21 plus existing bugfixes.
  
• Heat flux and Internal heat generation can be functions of both time and position with the CCX solver.
  
• Temperature solution variable for mechanical analysis types with CCX solver and static and DC current with internal solver.
  
• Reading CCX's .cel file for contact states is faster.
  
• Node rotation with an angle shows the angle in the outline tree like displacement does.
  
• .Net Framework version changed from 4.5 to 4.8
  
• Uniform general section’s “Perimeter length” property is enabled for mechanical analysis types to allow some load types.
  
• Removed vector magnitude option from Sum tool.
  
• Removed electrostatics from Static 2D analysis type. It’s still available in Static 3D.
  
• Bug fixes.
  

Sergio

Thanks for this update, the new simmetry bc for beams works really good, attached first test with a security cage

prop_design

"Symmetry boundary condition."

I haven't used Mecway in awhile (I think v19 was the last version I used, due to retirement). However, I still like to read the forum. I'm confused as to the 'new' symmetry boundary condition. I recall doing symmetry in Mecway previously. Can someone explain how the new version differs from previous versions, in this regard.

"CCX updated to 2.21 plus existing bugfixes."

Is a ccx.exe file part of the install now or do you still have to compile it yourself?

It's nice to see Mecway still going. I hope it continues to be a quality, low cost, FEA program for decades to come.

Thanks,

Anthony

Victor

Hi prop_design. Good to see you again!

Good point about symmetry. I don't like to add redundant features and this does look like one. In most cases, the new symmetry doesn't do anything you couldn't easily do with frictionless support or displacement and rotation constraints. However, frictionless support doesn't behave as a correct symmetry boundary condition on shells or beams that aren't normal to the symmetry plane, as in Sergio's example. Directly constraining rotational DOFs is sometimes impractical to set up on shells with CCX because the constraints have to be tangent to each individual element. So for the particular case of many curved shells at funny angles with CCX, the new symmetry is necessary.

Yes, ccx.exe has been included for a long time but still only with the Spooles solver so a lot of people replace it with a Pastix or Pardiso one for better performance.

prop_design

Thanks for clarifying Victor. Everything makes sense to me now. I recall being confused as to how axisymmetric worked in Mecway, because there wasn't a menu item for it. So this seems a lot more clear. Or how the user would expect. You are right, there were other workarounds. I believe your user manual use to mention how to do it. So this works better with CCX and in more cases, plus it's clearer for a new user. I recall my testing, in the past, proved what you are saying (that the workarounds provide good results).

I may have forgot about your ccx.exe file, now that you mention it. I had always been pointing Mecway to the version from the CCX website, so that you could use Pardiso. I guess you made bug fixes though, that aren't in the CCX website version? If so, I hope Guido will add them to the mainline. It would be really good if the version on the CCX website (that works with Pardiso) could somehow be included with Mecway. Or, even if you made your own branch that worked with Pardiso. Oh the joys of open source insanity.

Victor

Guido has implemented fixes to to most of the same bugs I did but just after the release of 2.21 so they're only in the Github source at the moment.

According to my understanding of GPL 2.0, nobody is supposed to redistribute CCX compiled with Pardiso, so I don't do that. Pastix may be fine, and seems reliable as long as you turn off mixed precision, so I hope to include that eventually.

prop_design

that's good to hear about the bug fixes.

the way the version on the ccx website works is it has spooles and pastix embedded and if you copy the pardiso dll files into the ccx.exe folder it will also run pardiso. it would be nice if the ccx.exe file you provide had the same functionality. that way you can use the solver= command with all three solvers. spooles is useless for me. it's slow and only works on very very small models.

prop_design

oh, i forgot. besides the methods in the old mecway manual, i also used cyclic symmetry to model axisymmetric. i also ran the full model. all of the models yielded the same results. it was for a solid model though. i never tried with shell or beam. also, maybe my boundary conditions were just fortunate to work in all ways.

cwharpe

Great having the options of N-S coupling for the internal solver! and Thanks for cleaning up the Sum Tool. In haste, I kept making the very error (Sum Magnitudes) I had cautioned about.

@prop_design: No Fair retiring before the rest of us. Hoping you will continue to contribute your valuable insight and experiences.

Sebastianmaklary

Awesome!
I'm especially happy with the inclusion of node-surface connections being added to the internal solver so the rigid spider elements can be retired from this use.

prop_design

@cwharpe

lol, thanks. It sounds crazy, but work as long as you can. Mecway has helped keep my mind active for many years. I don't think engineers can ever really retire. Even when your body stops, your mind keeps going. So many of my coworkers came back, after retiring. I always thought they were nuts. Now I get it.

MikeMcMullen

I get it too. Currently working about 1/4 time as a temp at the employer I retired from in 2003. Using new Mecway release to check the effect of a cosmetic change to a bridge rail.

Sergio

Have tested the new node-surface coupling features on the internal solver and they are great, is a big advantage for working with with real shell elements.





Have you thinked in adding no lineal materials to the internal solver as well?

Victor

Thanks for all the screenshots @Sergio . Always good to hear that features are useful or not. @Sebastianmaklary and @cwharpe too.

The internal nonlinear solver is waiting on ice for a time when I can add some useful alternatives to CCX, particularly beams and shells. I don't think just adding nonlinear materials to it will be very helpful otherwise. When it was enabled, I'd get questions from people where the answer was "You have to use CCX for that." so it was just a time-waster.

Sergio

Mecway solver has become better for shells and beams than CCX imho, maybe that extra no lineal material option could atract more users.

Whit this last upgrades such as the posibility to glue different meshes with TIE or couplings makes more suitable for solving assemblies. Maybe in the past one could say, why should I use the internal solver if there are some basics features that I need that are not available... but now? With the internal solver we can avoid that extra thinking and worries due to the expansion, the simmetry errors, knots on the corners, and all the explanations to our clients that those solid elements are equivalent to shells on other solvers, the internal has stress results at lower/middle/top surface like any other clasic solver, comparations are more direct and accurate than for CCX.

Not only that, with the internal solver we can see the bc on the results, that is also a great advantage for making nice presentations.

Victor

@Sergio I see what you mean now. The existing shells and beams don't work with nonlinear large displacements and upgrading them for that isn't trivial. That's the main work needed to make the internal nonlinear solver worthwhile, in my opinion.

prop_design

I'm with @Sergio on Mecway internal nonlinear solver coming back. Especially for shells and beams. The ccx shell expansion does cause issues, in model setup, a lot of times. I did find solids were better than shells though. shells don't capture the zz composite stress. shells assume this to be zero. so shells will show parts are ok when they will actually fail. this typically happens in fillets or curved surfaces. it wasn't something i was aware of, until i did a lot of testing of ccx shells. at first, i hated all the extra nodes, because it slows things down and limits the model size. it basically removes the benefit of using shells. but on the plus side, i found a failure mode i didn't know about previously. in any event, it would still be a big benefit to get the internal nonlinear solver back. i understand it's a lot of work and will take a lot of time. but i really believe it will be useful. especially if Coriolis effects are included for stress and modal analysis.

Andrea

I agree with Sergio. NL solver for pure shell elements or beams could be useful.
I notice that internal solver is more quckly on the latest versions...Is it true?

Victor

@Andrea version 20 had speed improvements for the internal solver.