Native Calculix windows version released by General Electric

svingegneria

Dear Mecway users,
at the following link GE repository with latest certified CalculiX Windows version:

https://github.com/GeneralElectric/CalculiX

Best Regards,
Stefano

Sergio

Stefano, excuse my ignorance, but can you point me exactly to the Windows executable file? I have download several of them and can't find any .exe

Regards!

kwip

Hi Sergio,

https://github.com/GeneralElectric/CalculiX/blob/master/releases/CalculiX-GE-OSS-2.10-win-x64.zip

unzip --> bin folder (ccx.exe and ccx_MT.exe as well as several dlls you might copy/move together with the exe-files)

regards

VMH

kwip,

Do you know the differences between "ccx.exe" and "ccx_MT.exe". And are these the same as the one you complied? Thanks

kwip

I did not try it so far, but I expect them to differ in the same way as my binaries.

svingegneria

Dear Sergio,
kwip answered greatly...sorry for delay!
Regards,
Stefano

VMH

FYI. I did a test to check out the solvers.

See attachments for a comparison among several CCX solvers: ccx2.8p2 complied by bConverged, ccx2.10_MT complied by Kwip, and ccx2.10 complied by General Electric (GE).

A cantilevered flat plate with dimensions of 1/2" x 24"x 36" loaded with 1 kip at the cantilevered end was selected for the test.

Component 1: Linear Quad Shell Elements (Quad 4)
Component 2: Quadratic Quad Shell Elements (Quad8)
Component 3: Quadratic TriShell Elements (Tri6)
Component 4: Quadratic Hex Solid Elements (Hex20)
Component 5: Quadratic Tet Solid Elements (Tet10)

From the results, ccx2.10 version from Kwip and GE has identical results. ccx2.8p2 and ccx2.10 versions had slightly different results for the shell elements.

Sergio

One thing that I found in the kwip version 2.10 is that when I ask for writing data in an external file (*NODE PRINT), if the value exced 9.9E+99 or 9.9E-99, it became 9.9+100 or 9.9-100 (so I has to edit by hands to plot in Excel)

Regards

rhoka

hi,

version 2.10 from bConverged has been publish by Jeff B., but only for 64bit. i'm still wait for my 32bit OS and i hope for his kindliness to build and provides.

http://www.bconverged.com/data/content/CalculiX_2_10_win_001.zip

greetings,

apadzak

I have installed 2.10 from bConverged and couldn't perform PipeClip analysis from Mecway's samples folder done in Mecway v4 and v5. Comming back to 2.8 bConverged made it working once again.

Is it just me?

Thanks

avico

It is not just you.

Linking to ccx.exe I have no result (ccx output in Mecway is empty) and linking to ccx.bat (in bin directory) I get error (divergence with values in second iteration like

average force= 12138039319532977555217129289274518421059859334854355145602367488.000000
time avg. forc= 12138039319532977555217129289274518421059859334854355145602367488.000000
largest residual force= 219404421944374718847865764582127078647926855741908179593220990173184.000000 in node 254 and dof 3
largest increment of disp= 8.811454e+18
largest correction to disp= 8.811454e+18 in node 254 and dof 3

Sergio

I can't solve the clip and others models that I have done before (on a fresh install of Windows 10, Mecway and CCX suite) on , even if I setup a bat file. I will back to the kwip version that runs smooth.

Regards

AndresLT

Sergio

I have the same problem with Calculix 2.10

Victor

Thanks everyone for bringing this up. I find it won't work or show any output if you use ccx.exe. Instead, use ccx.bat which sets some environment variables. With that, I was able to solve the pipeclip example. It might be because it depends on dlls that are in another folder.

kwip

the latter one is the reason why I've built ccx statically. Sergio, 9.9E99, 9.9E-99 ... you are pushing the limits infty vs. zero

Sergio

Kwip, you are right, but in my example of the snap belt I has those kind of results for loads, obviously they are zeroes, but I has to fix it by hand in order to plot on Excel. Who knows, maybe somebody test a nuclear reactor and fails due to this little numbers