XANSYS Message: 20691 [Go back to message list]
[bookmark on del.icio.us]
No rating yet
Rate item:

Subject: Re: Superelement/Substructuring
Author: Stephane Pierre Bordas spb276@
Date: 2001-01-31 12:18:00

Thank you very much Kimravn for your helpful answer, I apologize for
having posted the same mesage several times, but I did not see it appear
on the board and thought it was not sent. I am sorry it caused everyone
trouble.

I am sure I will figure out how ANSYS works and knowing that yourself did
get to the stifness matrix encourages me to dig further.

Concerning the FORTRAN application you are mentioning, I am not sure I
understand clearly what it exactly does. Could you please give me some
more details about it? We are indeed working on a UNIX system and the code
that will handle the use of the superelement stiffness and assemble it
with the mesh on which crack propagation will be handled is in C++.

I would be very interested in knowing more aobut the application you are
telling me about in your message as well as take a look at the sample
codes you are mentioning. I understand that you are not allowed to
disclose any more precise information about this code, but as you are
saying, I would be interested in being able to contact the individuals who
could send me the source code.

From what I am understanding, the biggest problem might come from the fact
that the stiffness file is in binary format. If I am getting things
right, the FORTRAN application that you wrote used the same binary format
for its input as ANSYS, which allowed you to get to the actual stiffness
values in the stiffness matrix and not just a criptic binary file. My goal
would be to become able to import this matrix and convert it to an array
variable in FORTRAN or C++ that I could then assemble with the rest of the
stiffness matrix (the one corresponding to the detailed model of the
subset of the structure I am studying). Do you see what I mean ?

Concerning the substructuring, as you must have understood, I am a
starting user, and I do not know the difference between substructuring and
superelement method. Isn't all this related to the static condensation
method ? In any case, it seems that from your experience, it be not the
easiest way to go, even if I did not exactly understood the trouble you
had when you tried to optimize this fillet you are mentioning.

To conclude, I would like to thank you all for your responses, any further
help will be greatly appreciated, and please, don't hesitate to let me
know if you think what I am undertaking is stupid and how I could remedy
my foolishness,

Thank you,

Stephane

On Wed, 31 Jan 2001 kimravn.jensen@a... wrote:

> Stephane,

> Neither I nor Mark should claim to know what you are able to do, but
> I'd say that you may be on the right track. I once created a FORTRAN
> application on a UNIX system which uses as its native binary format
> the same format that ANSYS uses on all platforms. This choice of
> software and hardware platform may be the reason why I remember the
> whole superelement stiffness retrieval thing as rather
> straightforward.

> I am at present not authorized to give you any code samples, but I
> could pass you the name of somebody who is. You could also start by
> asking your ASD for assistance.

> K.S. Raghavan's idea of using submodeling kind of scares me. I was
> once thoughtless enough to try to optimize a fillet as a submodel.
> The optimizer dug away the entire model - no steel, no stresses!

> Best regards
> Kim Ravn-Jensen

> Alfa Laval Copenhagen A/S - Maskinvej 5 - DK-2860 Søborg - Denmark
> Phone: +45 39536000 - Direct Phone: +45 39536511 - Fax: +45 39536560
> E-mail: kimravn.jensen@a...

> P.S.: Your final quote inspires me to pass you the motto of hardcore
> programmers: "Documentation is fiction, source is fact."

> --- In xansys@y..., Stephane Pierre Bordas wrote:
> > Hello,

> > For research purposes, my group is thinking of using ANSYS as a
> tool to
> > generate superelement stiffness matrices for structural stress
> analysis.

> > At this point, we have an I-DEAS model from which we can export a
> > universal file. There may also be a way in which we could export
> our model
> > to an ANSYS compatible format, but this needs to be checked.

> > Our goal is to isolate a given region of this model, and make a
> > superelement out of the remainder of the model, whose behavior we
> are not
> > interested in. Then, we would use this superelement in conjunction
> with
> > another, more refined mesh for the region of interest and run some
> crack
> > propagation studies with the code that was written in our
> department at
> > Northwestern University.

> > We don't know much about Ansys, but it seems to us that it would be
> the
> > way to go. How feasible does it seem to you ? How would you go
> about doing
> > it ? Has any of you already done it ? Where can I find an example
> of this
> > kind of study/the format of the superelement file ?
> > In particular, is it possible to access the stiffness matrix
> representing
> > the superelement ?

> > Thank you very much for your help,

> > Stephane

> > Stephane P. Bordas
> > Ph.D Student in Theoretical and Applied Mechanics
> > Northwestern University, Evanston, IL, U.S.A

> > http://www.tam.nwu.edu/sbordas/

> > "On ne doit appeler science que l'ensemble des recettes qui
> marchent, le
> > reste n'est que literature"

> > Paul Valery.

Stephane P. Bordas
Ph.D Student in Theoretical and Applied Mechanics
Northwestern University, Evanston, IL, U.S.A

http://www.tam.nwu.edu/sbordas/

"On ne doit appeler science que l'ensemble des recettes qui marchent, le
reste n'est que literature"

Paul Valery.

Posts possibly associated with message #20691AuthorDateScore
20654Superelement/SubstructuringStephane Pierre Bordas spb276@2001/01/30 
20656Re: Superelement/SubstructuringMartin Liddle2001/01/30 
20664Re: Superelement/SubstructuringKim Ravn-Jensen2001/01/31 
20691Re: Superelement/SubstructuringStephane Pierre Bordas spb276@2001/01/31 
20706Re: Superelement/SubstructuringDan Bohlen2001/01/31 
20739Re: Superelement/SubstructuringGerald Jones2001/02/01 
20960Re: Superelement/SubstructuringK.S. Raghavan2001/02/09 
25447Superelement/Substructuring Analysistap_club@2001/06/29