Re: [mor4ansys] Reduced-Order Models for the Calculation of Thermal Transients of Heat Conduction/Convection FE Models

[Panagiotis Koutsovasilis]

Hello Evgenii,

in answering your indirect query regarding any kind of comparison between the Arnoldi and the CMS method, please let me direct you my homepage:

http://tu-dresden.de/die_tu_dresden/fakultaeten/vkw/tgf/Itgf/mitarbeiter/panos

There you will find several contributions concerning the aforementioned matter. If  I am not mistaken, you had included, some time in the past, certain of these publications in some publication-list of the site http://modelreduction.com/.

Best Regards,
P. Koutsovasilis

2008/11/23 Evgenii Rudnyi <use...@rudnyi.ru>

Tamara has found an interesting reference and I got a reprint from the
authors. Some comments are below.

D. Botto, S. Zucca, and M. M. Gola, (2007) 'Reduced-Order Models for
the Calculation of Thermal Transients of Heat Conduction/Convection FE
Models', Journal of Thermal Stresses, 30:8, 819 – 839
http://dx.doi.org/10.1080/01495730701415806

The link to Prof Gola is
http://didattica.polito.it/pls/portal30/sviluppo.scheda_pers_swas.show?m=525

The paper presents an interesting engineering application that
requires reduced thermal models. The authors have used the Guyan
reduction and the component mode synthesis to reduce the models.
Moreover the fluid part first has been statically reduced to leave the
transient behavior in the solid part only.

The engineering problem concerns nuclear and aeronautical applications
where the change in temperature causes thermal stresses and hence
fatigue. Online calculation of temperatures is needed to assess
fatigue damage accumulation and residual life.

My first observation was that the background mathematical problem in
the paper is very similar to that in electronics for cooling of chips.
There they need to estimate the temperature distribution in a chip
because when the chip temperature is too high, the chip fails. Clearly
the damage mechanisms in a chip are more complicated as pure fatigue
in a mechanical structure. Yet, the part related to thermal
engineering is pretty similar. See for example papers for the PROFIT
project

http://www.hitech-projects.com/euprojects/profit/

The second thought was that this would be an interesting application
for MOR for ANSYS. We have compared Arnoldi with Guyan in [1]. I have
not seen the comparison Arnoldi with CMS yet. It would be interesting
to make it. Well, I should confess that I have not yet understood how
many inputs will be necessary to introduce to treat the model in
question. Guyan and CMS are input independent and in this case this
does not matter. For Arnoldi the right answer is crucial for success.
Anyway for Arnoldi the unsymmetric heat conductivity matrix is not a
problem. It is working in this case the same way [2].

Evgenii

P.S. If you find interesting papers, please make your comments to the
list or just send me and I will make my comments.

[1] T. Bechtold, E. B. Rudnyi and J. G. Korvink, Automatic Generation
of Compact Electro-Thermal Models for Semiconductor Devices. IEICE
Transactions on Electronics, 2003, v. E86C, N 3, pp. 459 - 465.
http://search.ieice.org/2003/files/e000c03.htm#e86-c,3,459
Preprint at http://modelreduction.com/doc/papers/bechtold03IEICE.pdf

[2] C. Moosmann, E. B. Rudnyi, A. Greiner, J. G. Korvink, Model Order
Reduction for Linear Convective Thermal Flow. THERMINIC 2004, 10th
International Workshop on Thermal Investigations of ICs and Systems,
29 September - 1 October 2004, Sophia Antipolis, France, p. 317-321.
http://modelreduction.com/doc/papers/moosmann04THERMINIC.pdf

[Evgenii Rudnyi]

Hi Panos,

> in answering your indirect query regarding any kind of comparison between
> the Arnoldi and the CMS method, please let me direct you my homepage:
>
> http://tu-dresden.de/die_tu_dresden/fakultaeten/vkw/tgf/Itgf/mitarbeiter/panos
>
> There you will find several contributions concerning the aforementioned
> matter. If I am not mistaken, you had included, some time in the past,
> certain of these publications in some publication-list of the site
> http://modelreduction.com/.

Thanks. I have not forgotten your works. I have just meant a thermal
problem. The behavior of a thermal and structural problems are
different. Say, mode superposition is working for structural mechanics
but does not work for a thermal problem. Hence it would be good to make
comparisons separately.

I see that you make good progress. When you publish a paper, please send
a note to the list. It will be interesting for others.

By the way, could you please send me reprints of

Multibody System Dynamics, 20(2), 111-128 (2008)
Mathematical and Computer Modelling of Dynamical Systems, 2008.

I would appreciate it.

Best wishes,

Evgenii

[Panagiotis Koutsovasilis]

Hello Evgenni,

I would be glad to send you the reprints. I can do it, though, during the week that follows, because I am at the moment out of office.

Best regards,
Panos

2008/11/24 Evgenii Rudnyi <use...@rudnyi.ru>