[Inquisitor]

I been writing a C#.NET application to help me generate the Z88 model input files. And I've noted that the analysis programs seem to be separate binaries in the C:\Z88AuroraV2\win\bin directory. Are these callable so that I can run the analyses on the input files my app creates? If so, is there some documentation for the command line parameters necessary?

Thanks.

[mifris]

Hello Inquisitor,

the .exes are always binary. If you need some sources then i recommend you to download z88v14os. The input-files, which were used there can be imported in Z88aurora. This program is open source, written in C.

Best regards

M. Frisch

[mz15]

Hello Inquisitor,

yes the modules are callable, but not all are documented. You can call the the Solver with the Parameres are documented in the theory-manual page 101. All other converters have specific parameters depend on the current element, and so on. Therefore, these are not documented separately.
If you call the exe without any parameter you can see a list what parameter the executable take.

regards
Markus Zimmermann

[Inquisitor]

I have my program creating these files:

• z88control.txt
• z88structure.txt
• z88setsactive.txt
• z88sets.txt

If I start up Z88Aurora and open the project directory, I can see the generated model and successfully run a linear-mechanical and non-linear mechanical analysis, so I think I have generated the files necessary. I next want to run the solvers without using the Z88Aurora GUI, so...

1. I then opened up a standard command prompt
2. Changed directory to the project folder
3. I then ran: C:\Z88AuroraV2\win\bin\z88r -t –parao
4. I get: ### cannot open Z88.DYN ..stop ###
5. I found the z88.dyn file in the bin folder and copied it to my project folder and re-tried (3)
6. I then get: ### cannot open Z88MAN.TXT ..stop ###

Basically, I gather, I need to convert the four original files above to files that z88V14 OS needs. Is there some documentation or could you list the applications I should be using? I gather that some of the programs needed and parameters might be based on things in my model. If so, here are the basics…

1. Relatively small models ~1000 dof
2. Trusses and/or Beams
3. PARDISO solver
4. linear and/or non-linear mechanical (at least on the trusses)

Thank you for your help,
Inquisitor

[ccad]

Hello Inquisitor,

basically, Z88R from Z88Aurora needs the same input files as Z88R from the open source Z88 V14, i.e.
- Z88I1.TXT (very similar to Z88STRUCTURE.TXT)
- Z88I2.TXT (the boundary conditions)
- Z88I5.TXT (the surface loads)
- Z88ELP.TXT (the element parameters)
- Z88INT.TXT (the integration orders)
- Z88MAN.TXT (some flags for running the solver, for example the beam flag IBFLAG)
- Z88MAT.TXT (the materials, pointing to the .TXT files, see next line)
- one or more .TXT files for the materials
- Z88.DYN (the memory needs)
Please see the Z88 V14 OS (for Open Source) documentation. But - why using the Z88R solver from Z88Aurora? For a structure with less than 1,000 DOF with beams and trusses, the open source solver Z88R with the Cholesky option will do best.

Best regards

Prof. Rieg

[Inquisitor]

Hello Inquisitor,
... But - why using the Z88R solver from Z88Aurora? For a structure with less than 1,000 DOF with beams and trusses, the open source solver Z88R with the Cholesky option will do best.
Best regards

Prof. Rieg

Hello Professor Rieg,

Thank you for your quick response. It has been a long time since I've been in deep with FEA and I have only been using your software for about a month and only during my "free" time. I should add... it has been both enjoyable and rewarding time. However... retaining everything I've read is not as easy as it once was. I will gladly take your advice on the Cholesky. Also, I has assumed that the Z88R solver was identical in both the OS and the Aurora versions. I will start digging into the OS version documentation.

As a side note, I would like to make my program available on your forum in case it offered some benefit to other users and I didn't want to restrict it to my minor needs. Who knows?... maybe one of your users might find it useful for a million beam model.

vbr
Inquisitor

[ccad]

Hello Inqusitor,

yes, the two linear solvers Z88R (Open Source Z88 and Z88Aurora) are quite identical - i know it quite exactly because i wrote them both. Here are the differences:

- the Z88Aurora solver Z88R features with the -parao option multi-CPU support (this is only of interest for large 3D structures with tetrahedrons or hexahedrons but not at all for beams & trusses structures). By the way, multi-CPU support for the iteration solvers SICCG & SORCG is comming soon - with the next service release - i guess, in the near of X-mas this year.

- the Z88Aurora solver Z88R features automatic memory allocation by running first Z88R with the -t (i.e. test) option filing a memory definition file Z88R.DYN which is read in by a second run of Z88R with the -c (i.e. compute) option. However, the Z88 Open Source solver Z88R specifies the memory needs in the file Z88.DYN (which has more entries than Z88.DYN of Z88Aurora); thus, you must specify the memory needs in advance which is less comfortable for the user.

Of course, we will appreciate your solution to be offered on our internet pages - thanks a lot in advance!

Best regards

Prof. Rieg

[Inquisitor]

Professor Rieg,

Summary: I have my application successfully creating the geometry and running static linear analyses, but I’m having troubles starting a non-linear analysis. Here’s what I am doing so far.

I generate:

• Z88Man.txt
• Z88.dyn
• Z88I1.txt
• Z88I2.txt
• Z88Mat.txt
• Z88Elp.txt
• Z88Int.txt

I then call executable in the C:\Z88AuroraV2\win\bin folder.

• z88r –t –choly
• z88r –c –choly

I can then read back and process the output files

• z88o2.txt
• z88o3.txt
• z88o4.txt

This is all working fine.
Now for a non-linear analysis, I tried running z88nl on the same files from above. Per the instructions in 4.6 of the Theory Manual, I also added the zero’d Z88I5.txt and Z88NLI7.txt and added the NONLINEAR START block of statements to the Z88MAN.txt file. But it gives me an error saying the Z88MAN.TXT file is wrong. Below, are my Z88MAN.txt files and the results from the z88nl.exe. Would you be able to point me in the right direction?

Regards,
Inquisitor

Code: Alles auswählen

DYNAMIC START
  GLOBAL START
    IBFLAG 0
    IPFLAG 0
    IHFLAG 0
    SIMCASE 1
  GLOBAL END
  
  SOLVER START
    MAXIT 10000
    EPS 1e-007
    RALPHA 0.0001
    ROMEGA 1.1
  SOLVER END
 
  STRESS START
    KDFLAG 0
    ISFLAG 1
  STRESS END

  NONLINEAR START
    NLFLAG                   1
    NLAERH                  25
    MAXNLIT               1000
    EXIT                     1
    TOL                   1E-7
    AUTOGAUSS                0
    OUTPUTFLAG               1
    OUT_CAUCHY               1
    OUT_INT9OFFS             1
    PARSP                    1
    BGLAENG                1.0
  NONLINEAR END
DYNAMIC END 

Code: Alles auswählen

                *****   ***    ***
                   *   *   *  *   *
                  *     ***    ***
                 *     *   *  *   *
                *****   ***    ***
   by Prof.Dr. FRANK RIEG (C), Germany, 2011
Z88NL the nonlinear solver for Windows, LINUX und Mac OS X
Univ.Prof.Dr.-Ing. FRANK RIEG and Dipl.-Ing. Christoph Wehmann
Part 1/5 of memory allocated.
Finished reading Z88I1.TXT.
Part 2/5 of memory allocated.
Part 3/5 of memory allocated.
*** Reading Z88Aurora V2 material data file 1.txt ***
*** Reading Z88Aurora V2 material data file 2.txt ***
*** Reading Z88Aurora V2 material data file 1.txt ***
*** Reading Z88Aurora V2 material data file 2.txt ***
*** Reading Z88Aurora V2 material data file 1.txt ***
Finished reading Z88MAT.TXT.
Finished reading Z88INT.TXT.
Finished reading Z88ELP.TXT.
Part z88ai finished successfully.
Saving only non-zero elements: +5.2400E-003  MB for KT.
Saving matrix completely: +3.4848E-002  MB for KT.
kfoun: 654.
Part z88a1nl finished successfully.
Reading file Z88I5.TXT finished.
Reading file Z88I2.TXT finished.
Part z88a2nl finished successfully.
Part 4/5 of memory allocated.
### Z88MAN.TXT is wrong ..stop ###

[mz15]

Hello

maybe there are some parameter missing. If you look into a z88man.txt while an non-linear analysis runs in Z88Aurora V2 there are more parameter in the file. The parameter below are from the created when run the example 32.


NONLINEAR START
NLFLAG 1
MODN_STEP 1
NLAERH 5
MAXNLIT 1000
EXIT 1
TOL +1.0000000E-007
DIFF_FLAG 0
UPDATE 0
DELTA +1.0000000E-004
SFLAG 1
AUTOGAUSS 0
PLOTFLAG 0
OUTPUTFLAG 1
OUT_DEFGRAD 0
OUT_LAMBDA 0
OUT_PK1S 0
OUT_PK2S 0
OUT_CAUCHY 1
OUT_CMAT 0
OUT_EPSILON 0
OUT_ENERGIE 0
OUT_GESKRAEFTE 0
OUT_INT9OFFS 1
OUT_INT12OFFS 0
OUT_INT36OFFS 0
PARSP 1
ELKORR 2
DELTA_ANZ_HW +1.0000000E-004
ELEFORM 1
BGLAENG +1.0000000E+000
LASTST 0
NONLINEAR END

maybe that helps you to run your example.

regards
Markus Zimmermann

[Inquisitor]

Oh... the life of graduate school. I remember "fondly" having reviews with my professor at 2:00 AM.

And... its interesting how aging turns things like that into fondly.

Thanks Markus. I'll give that a try when I get home from work this evening.

Inquisitor

[Inquisitor]

Thank you Markus...

That's what I've been doing most of the time. The linear static analysis always left the files behind after closing Z88Aurora. When I ran the non-linear case, I never found the Z88Man.txt file.

...until I left the "Z88NL: ran successfully!" window up... per your suggestion... and now I found all kinds of interesting files!

[mz15]

Hello,

This pleases me. I hope you will now continue with your calculations.

regards
Markus Zimmermann