Coupling with *Distributing in 2.19

1

First of all, the new algorithm shows a load distribution which is compatible with a linear underlying stress distribution. This should mitigate problems with the singularity of the hyperbolic distribution in earlier versions.

Note that the new algorithm is rather a pre-processing feature than a multi-point constraint, which is fine for small displacements.

I observed a nasty problem: In my tests with 2.19 initially *distributing didn’t work at all. However, the official example coupling7.inp worked and so did a minimum example with a single C3D8I element.

It turned out that the problem was in the *cload card, which didn’t seem to accept nset identifiers any more.
The following example has two steps

  1. cload applied to node 9
  2. cload applied to nset Nload (which consists of just node 9)
  3. cload applied to a set of ordinary nodes (not being reference of a coupling)

Only in step 1 and 3 the load is actually applied. This means that *cload usually accepts nsets as node identifiers, but not if these nodes include reference nodes of a coupling card. Can anyone reproduce this problem?

*NODE, NSET=Nall
    1,-1.0, -1.0, -1.0
    2, 1.0, -1.0, -1.0
    3, 1.0,  1.0, -1.0
    4,-1.0,  1.0, -1.0
    5,-1.0, -1.0,  1.0
    6, 1.0, -1.0,  1.0
    7, 1.0,  1.0,  1.0
    8,-1.0,  1.0,  1.0
    9, 0.0,  0.0,  1.0
*ELEMENT, TYPE=C3D8I, ELSET=Eall
     1,     1,     2,     3,     4,     5,     6,     7,     8
*NSET,NSET=Nfix 
1, 2, 3, 4
*NSET,NSET=Nload
9
*nset,nset=ntop
5,6,7,8
*SURFACE, NAME=Sload
1, S2
*boundary
Nfix,1,3
** material definition
*material, name=dummy
*elastic
10000,0
** material assignment to bodies
*solid section, elset=Eall, material=dummy
** coupling
*coupling,surface=Sload,refnode=9,constraint name=load
*distributing
1,6
** Moment about x
*step
*static
*cload
9,4,1000.
*node file
U
*el file
S
*section print, surface=Sload, name=so1
SOF
*el print, elset=Eall, totals=only
ELSE
*end step
*step
*static
*cload, op=new
Nload,5,1000.
*node file
U
*el file
S
*section print, surface=Sload, name=so1
SOF
*el print, elset=Eall, totals=only
ELSE
*end step
*step
*static
*cload, op=new
ntop,3,1000.
*node file
U
*el file
S
*section print, surface=Sload, name=so1
SOF
*el print, elset=Eall, totals=only
ELSE
*end step

(Edited to add the missing ntop definition)

1 Like
2

I recall a similar issue with not being able to apply loads to nsets. I will give it a try later today and report back.

3

I tried running it in 2.19 and I got the following error message:

*ERROR reading *CLOAD: node set NTOP has not yet been defined

1 Like
4

You are right, I forgot to paste the ntop definition when adding the 3rd step while editing the post :frowning: Yet this only affects the ordinary load application, not the coupling. The correct input file is found at github.

5

I checked this input file and I can confirm that the load is applied only in the first and third step. The second step results in zero forces, stresses and displacements. Should we notify Dr Dhondt ?

6

Hello Mr. Kraska,
could there be a little discrepancy with the index of the load distribution of torsion of the shear stress?

image

wbr

7

yes i can confirm version 2.19 shown no results, but is not issue with version 2.17

CCX2.17-CalculiX GraphiX
CCX2.17-CalculiX GraphiX772×751 16.4 KB

1 Like
8

You are right, shear stress is caused by Mz. Fixed.

9

I’d propose to use the issues feature of the calculix-archive github repository. This should be the most official place available for discussion.

In order not to spam the developers, we could agree on always first discuss issues here and only if they are confirmed, to convey them there.

10

Do you mean this place: https://github.com/calculix/ccx ?
I was thinking about reporting bugs on github but does any dev actually look there ? Also, the repository seems abandoned.

11

Sorry for being unspecific, I mean this: https://github.com/Dhondtguido/CalculiXarchive

This could be better than here, because from the developer’s point of view, a filtered issue list would be less noise than here, where the focus is on mutual user support.

People sometimes post issues in my example repository, yet I don’t seem to get email notifications from there. Probably a settings problem. Email notification is essential for fast response, because at times I am not at github for months.

2 Likes
12

Thanks, I didn’t know about this repository. I thought that the current source code is available only here: http://www.dhondt.de/
Indeed, this repo looks like the best place to report bugs and maybe also request features. I hope that Dr Dhondt doesn’t mind.

13

According to the release notes for ccx 2.20, *Coupling in combination with *Distributing was recoded. However, I checked and this bug still persists :frowning:

14

Hi,

can anyone send me the correct input deck to dhondt@t-online.de. I would like to investigate this problem (Sept 2020).

Best Greetings and thanks,

Guido

15

I meant Sept 2022…
Guido

16

I can send it, I was going to write an e-mail to you anyway.

17

@dhondt I have sent you an e-mail with this input file. I am writing this just to let you know in case it goes to spam or something and to let others know that the file was sent.

18

I can confirm that this bug is fixed in CalculiX 2.21.

2 Likes