New CalculiX discussion group

Victor

Guido (the author of CalculiX) set this up and participates there. It replaces the Yahoo! group:

https://calculix.discourse.group/

prop_design

thanks victor,

there was a useful post there about version 2.17. guido is saying the PaStiX solver is up to 4x faster than the Intel MKL Pardiso solver. also, if you have nVidia graphics card with CUDA support that can go up to 8x faster. not a lot of details, but sounds promising. Also, he mentions that the Mortar contact isn't ready for prime time yet.

he mentions you need PaStiX4CalculiX on Github. if this can be distributed, then maybe we can finally have an easier Mecway install. no longer having to install and setup the Intel MKL.

there is more info on the CCX website. it looks like right now the install for PaStiX is difficult. you have to download and compile a bunch of stuff. Also, that process still seems to rely on the Intel MKL to some extent.

pschwenn

Wing Structure Question

(This question likely belongs elsewhere - make a suggestion as to where)

In a cgx display of semi-monocoque wing forces, since the skins must be unloading stresses to the ribs (and thence to the spars), one would expect that those stresses would take a discontinuous jump along the wing skins at the junctures of the ribs (and of course not to do so with a pure monocoque wing without ribs.)

But in the ccx+cgx calculation/display of a simple wing drawn and meshed in OpenVSP, the stress (Worst PS or Mises or Tresca) there's no great difference between the semi-monocoque and pure monocoque cases, even at high G-loads. See attached (semi-monocoque first image, monocoque second).

Any ideas?

Thank you

Victor

Hello pschwenn. Best to try the CCX group https://calculix.discourse.group/

But since you're here, a couple of general ideas to bear in mind:

A solution that looks continuous is consistent with a discontinuous one for small enough discontinuities. If you estimate the expected difference in stress on either side of a rib, you'll know if you can expect it to be visible on the plot or buried in discretization error.

The .frd file format (and CGX, think) can't represent discontinuous stress in a connected mesh because it only stores values at the nodes and discontinuities get averaged out. So you may need a finer mesh to see them.