Das Forum zu Z88OS (Open Source) / The newsgroup for Z88OS (open source)

Moderatoren: ccad, mz15, auroraIco, Lehrstuhl

vom TomGroves
#2725
I'm trying to use z88r on 2D quadrilateral meshes (from the command line, because it's going to be called from another program). I'm using a unix64 machine, and my installation of z88r runs quite happily on example b1 (either with -choly or -siccg), which is the closest to what I'm trying to do.

However, when I run
Code: Alles auswählen
z88r -c -siccg
on my own input files, produces
Code: Alles auswählen
*** SICCG solver with sparse matrix storing ***
*** computing linear structure ***
reading solver data file Z88MAN.TXT
reading structure data file Z88I1.TXT
reading coordinates
reading element informations
reading integration order data file Z88INT.TXT
reading element parameters data file Z88ELP.TXT
reading material data file Z88MAT.TXT
>>> start Z88AI <<<
program allocated 191 MB memory
still       942 steps, pointer IEZ =         1
and then simply stops.
The z88r.log ends with
Code: Alles auswählen
start reading function RI188I
reading Z88I1.TXT
reading coordinates
reading element information
Z88I1.TXT read, leaving reading function RI188R
start Z88AI
Using the Cholesky solver is a little better, but still doesn't work:
Code: Alles auswählen
*** Cholesky solver with Jennings storing ***
*** computing linear structure ***
reading solver data file Z88MAN.TXT
reading structure data file Z88I1.TXT
reading coordinates
reading element informations
reading integration order data file Z88INT.TXT
reading element parameters data file Z88ELP.TXT
reading material data file Z88MAT.TXT
>>> start Z88A <<<
no. 182 type 7                        
writing Z88O0.TXT
vector GS needs 1436523 elements
vector KOI needs 1468 elements
>>> start Z88BR <<<
program allocated 92 MB memory
no. 182 type 7                        
>>> start Z88CR <<<
reading constraint file Z88I2.TXT
incorporating constraints pass 1
constraint no. 979 type 1incorporating constraints pass 2
writing Z88O1.TXT
start SCAL88

### diagonal element in array zero or negative..stop ###
### often caused by missing or wrong constraints     ###
### recover: check constraints (underdefined ?)      ###
My program produces z88i1.txt, z88i2.txt, z882mat.txt, z88elp.txt and z88int.txt. All other input files are the same as in example b1.

All the elements are type7, 8-point quads, simulating 1mm-thick steel sheet.

I can view the structure in z88o, which displays the wireframe and boundary conditions correctly.

z88mat.txt:
1
1 182 51.txt

z88int.txt:
1
1 182 3 3

z88elp.txt:
1
1 182 1.000000 0 0 0 0 0 0

z88i1.txt:
2 943 182 1850 0
1 2 3.738256 -2.000000
2 2 3.083860 -3.084800
3 0 2.000000 -3.751697
...
1 7
114 142 162 128 216 217 218 219
2 7
94 149 86 66 220 221 222 223
3 7
131 117 129 169 224 225 226 227
...

z88i2.txt:
979
1 1 1 0.032369
2 1 1 0.061532
3 0 2 0.000000
3 1 2 0.000000
3 1 1 0.038771
4 0 2 0.000000
4 1 2 0.000000
4 1 1 0.022045
...

The object being simulated has lots of zero-displacement points at the bottom, and forces on all the points simulating gravity, so it shouldn't be underconstrained.

Could anyone offer suggestions as to what I might be doing wrong? I can PM the files if people want to try running it themselves, but the forum won't let me upload .txt attachments.

Thanks!
Tom
Benutzeravatar
vom SHautsch
#2729
Dear TomGroves,

we also discovered some wrong input parameters:

z88i1.txt:
2 943 182 1850 0
1 2 3.738256 -2.000000
2 2 3.083860 -3.084800
3 0 2.000000 -3.751697
...

z88i2.txt:
979
1 1 1 0.032369
2 1 1 0.061532
3 0 2 0.000000
3 1 2 0.000000
3 1 1 0.038771
4 0 2 0.000000
4 1 2 0.000000
4 1 1 0.022045

You have to specify the number of DOF per node in the z88i1.txt. Zero nodes is not allowed.
You also used DOF #0 in the z88i2.txt, which does not exist. And you have some double constraints applied at node 3 DOF #1 - displacement and force at the same time. Was this intended?
And you need to calculate the degrees of freedom from the number of nodes and dof per node, which should be 1886 in your case.

Please attach your project data ZIPPED, so we can have a detailed look at it.

Kind regards,
SHautsch
vom TomGroves
#2730
Wow, quick replies! Thanks :)
Looks like a Memory Problem. Did you adjust the file z88.dyn? Please See the proper section in the manual.
No, because the section on z88.dyn said it would tell me if it ran out of memory. I have now tried increasing several of the parameters, but it made no difference. I have included z88.dyn in the attached .zip file.
we also discovered some wrong input parameters:

z88i1.txt:
2 943 182 1850 0
1 2 3.738256 -2.000000
2 2 3.083860 -3.084800
3 0 2.000000 -3.751697
...
You have to specify the number of DOF per node in the z88i1.txt. Zero nodes is not allowed.
The nodes which have zero DOF are the nodes which have constraints specifying that they can't move. I thought that this meant that they have zero DOF. Do they retain full (ie 2) DOF even when fixed?
Also, I tried altering my program to let fixed nodes retain 2 DOF, and it made no difference.
z88i2.txt:
979
1 1 1 0.032369
2 1 1 0.061532
3 0 2 0.000000
3 1 2 0.000000
3 1 1 0.038771
4 0 2 0.000000
4 1 2 0.000000
4 1 1 0.022045

You also used DOF #0 in the z88i2.txt, which does not exist.
Thank you! Yes, I was counting from zero instead of from 1 :oops:
And you have some double constraints applied at node 3 DOF #1 - displacement and force at the same time. Was this intended?
Not really, but I tried doing the same thing to the wrench in example b1 and it ran quite happily. I can remove the gravity forces from the fixed point it you think it's a problem.
Please attach your project data ZIPPED, so we can have a detailed look at it.
Oh yes. :oops: The obvious solution!

Thanks again for the quick responses!
Dateianhänge
(14.31 KiB) 132-mal heruntergeladen
Benutzeravatar
vom SHautsch
#2731
Dear Tom,

the z88i1.txt contains the information about nodes and elements, that's your structure file. The element no. 7 is defined to have 2 DOF per node, so you have to type in this number in the node list. It's totally independent from your boundary conditions / z88i2.txt.

In the z88i2.txt, where you write down your loads / constraints, you tell the solver which DOF is fixed. The DOF then still exists, but the displacement at the given node in the specified direction is set to zero. You must not "delete" DOF, but set them to zero.

If you put a load and a displacement constraint on the same node, this should not be a problem for the solver. I just asked if there was an error in numbering or if that was intended. However this is additional work for you, as the force will not have any effect on the structure.

The main problem is that your structure is invalid. The middle nodes are overlapping, that means there are two nodes where there should only be one (see screenshot).
overlapping middle nodes
(167.45 KiB) 448-mal heruntergeladen
This cannot work. Please keep that in mind and have a look at your mesh generator. You could use our superelement mesher to perform a mesh update and make your structure valid (this would mean you will have to apply all constraints again) or tweak your mesh generator to produce a valid mesh.

Kind regards,
SHautsch

P.S. Zipped files reduce our server load and disk usage, text/ascii files are just too inefficient - so we don't allow them to be uploaded :mrgreen:
vom TomGroves
#2732
OK, I've sorted out the DOF problem. Thanks

I've also kept track of which nodes are fixed, and not bothered exerting force on them.
The main problem is that your structure is invalid. The middle nodes are overlapping, that means there are two nodes where there should only be one (see screenshot).
That's embarassing :oops:
In my defence, the reason I didn't spot that is that z88o can't load fonts on my system (says "Cannot load Pango Font! STOP!" at startup, with the CLI giving the error "GdkGLExt-WARNING **: cannot load PangoFont") so it can't display any labels over the wireframe of the structure :(

Anyway, I've fixed the problem in my mesh generator, I think. The node count dropped by about 1/3, which suggests the duplicate nodes have been merged.
I've installed a CAD package which can view .dxf files, and used z88x to convert z88i1.txt to z88x.dxf. The results look good in the CAD program - no duplicate nodes.
However this hasn't made any difference at all to the results when running z88r :(

Thanks again for all your help with this.
Tom
Dateianhänge
(15.37 KiB) 104-mal heruntergeladen
Benutzeravatar
vom SHautsch
#2736
Dear Tom,

your structure looks much better now. But there are still a few errors (see screenshot).
free nodes in structure
(204.92 KiB) 421-mal heruntergeladen
Your structure must not contain "free" nodes which are not connected to the rest of the structure.

Which OS do you use? Maybe we can work around that pango problem, so you can display labels on your machine without the need to use the dxf-conversion.

Kind regards,
SHautsch
vom TomGroves
#2739
Your structure must not contain "free" nodes which are not connected to the rest of the structure.
Aha! :D That's done it! Once I get rid of those floating points, it works!
And for once it wasn't my code which was to blame - for some reason GMSH was deciding that it didn't like certain quads, and dropping them from the mesh, but still returning the points which they were made from.

I'm using Ubuntu 12.04 64-bit. If there's a known fix for Pango, then please do let me know, but otherwise don't worry about it, because I can still use z88o to view the stresses. And I'll mostly be feeding the z88o*.txt files into another program anyway - I'm just using z88o for diagnostics.

Thank you so much for helping me with this! :thumbup:
Benutzeravatar
vom SHautsch
#2741
Dear Tom,

we are glad we could help you.

The font problem is known and results from changed X11 files, which were introduced with the newer versions of unix OS like Ubuntu. In fact this isn't a problem of Z88OS, but of the messed up X11 libs. We are waiting for a solution from the unix devs, but they don't seem to put a lot of work into that :roll:
For Z88Aurora, we use another way to display fonts and thus you could use Z88Aurora for displaying your results (or in fact for your whole calculations). You can run most of the executables of Z88Aurora from the command line and also convert the results to a Z88Aurora compatible format (have a look at z88ro2ag).

If you have questions on how to use Z88Aurora for your purposes, let us know.

For now, we consider your problem solved. Have fun with Z88 and get back to us if something is not working.

Kind regards,
SHautsch
vom sebastienl7
#3030
Hi,
I also use kind of Ubuntu based OS and I have "GdkGLExt-WARNING **: cannot load PangoFont" error message when I launch Z88O.

Nothing worked first in Z88, because the current working directory . is not part of PATH by default in linux.
I fixed this with
PATH=$PATH:.

Now I can see the wrench, I can move it with the mouse and everything seems to work fine except for the labels. They are not visible.

Does a solution exist for that problem?

Thank you for your help.
vom MBraun
#3032
Hi sebatienl7,

this is a ubuntu based problem. Because of a library update the labels aren't visible any more.
We are working on the solution!

Best regards,
Mbraun
Start

Hallo Schlorchi00, leider konnten wir den Fehler[…]

Hallo liebes Z88 Team, gibt es bereits Neuigkeite[…]

UPDATE Nachdem ich es nach weiteren Tests leide[…]

We're glad you got it working. We will have a look[…]