Welcome to the forums. Please post in English or French.

You are not logged in.

#1 Salome-Meca usage » max_base error in Code_Aster 15.4? » 2021-11-01 04:32:16

DougAtMcGill
Replies: 1

While running a very long DYNA_NON_LINE analysis in two steps I get the following error after about 12 hours:

<EXCEPTION> <JEVEUX_42>                                                                       

Fichier saturé, le nombre maximum d'enregistrement 61440 de la base GLOBALE est atteint
il faut relancer le calcul en passant une taille maximum de base sur la ligne de commande
argument "--max_base" suivi de la valeur en Mo. 

<EXCEPTION> <CATAMESS_55>                                                                     

Recursive calls of error messages or warnings.                                             

EXECUTION_CODE_ASTER_EXIT_10723=134

<I>_EXIT_CODE = 134

I tried adding the line this line to my .export file:
A args -max_base 300000

but still got the same error.

Another forum post mentioned that as of Code_Aster version 14, it should no longer be necessary increase the max_base argument and that instead the program will do it automatically. When I try running as_run with a "-max_base" argument it doesn't seem to accept it. I am using Code_Aster version 15.4 that comes packaged with Salome Meca in a Singularity container, if that makes any difference.

Does anyone have any ideas about how to fix this error? It is very difficult to troubleshoot when each attempted model run takes 12 hours to reproduce the error. I'm relatively new to Code_Aster and completely new to Singularity.

#2 Re: Code_Aster usage » Multi-step nonlinear dynamic analysis that keeps velocity.. [SOLVED]? » 2021-11-01 04:17:42

Hanbin wrote:

Maybe you can have a look at ETAT_INIT?  I think ETAT_INIT(EVOL_NOLI=LASTSTEP) can do that with EXIT=(newboundry=xxx) in your new DYNA_NON_LINE.Or I think you can use this with VITE=xxx , maybe xxx= CREA_CHAMP(the velocity you get in the first step)? Hope some other people can give better answers.

Thank you Hanbin. The EVOL_NOLI seems to work. I had read somewhere else that Code_Aster could only carry displacement between different analysis steps, but it certainly seems to run using ETAT_INIT as you suggested. I presume this is correctly carrying the velocity and acceleration from the end of the first analysis step into the second. I will confirm this at some point after solving other problems with the model.

#3 Code_Aster usage » Multi-step nonlinear dynamic analysis that keeps velocity.. [SOLVED]? » 2021-10-19 22:32:08

DougAtMcGill
Replies: 3

I'm trying to model a mallet head striking a bar using 3D solid elements. The bar is initially at rest and the mallet head is given an initial velocity. Collision between the mallet head and bar is captured using a contact surface and large displacements. After the collision the mallet head bounces away from the bar. At this point in the simulation there are no more forces on the mallet head, so it keeps moving in a straight line until the simulation ends. My goal is to capture the bar's displacement both during the collision and for some time after (where it essentially becomes free vibration with damping).

My question is: is there a way to eliminate the mallet head from the calculation after the collision has finished? At that point I no longer care what happens to the mallet head. However, since it keeps moving in a continuous straight line, its nodal displacements keep growing and growing (which, I imagine, could lead to some numeral problems or otherwise slow down the calculation).

I was thinking perhaps there might be a way to define a multi-step nonlinear dynamic analysis? The first step could capture the collision. Then, after a fixed amount of time, when I can be sure the mallet head is no longer in contact with the bar, I could define a second analysis step. In the second step I could add a new boundary condition that sets the mallet head displacement to zero, while allowing the bar to continue to vibrate. For this to work, the displacements, velocities and accelerations of each node in the bar would need to carry over from Step 1. A document called "Non-linear analysis - Advanced topics" from the Code_Aster website mentioned that displacements, stresses and internal state variables can be carried over between analysis steps, but it doesn't mention velocities or accelerations.

Is it possible to carry over velocities and accelerations between multiple steps of DYNA_NON_LINE analyses and model the mallet/bar impact as I've described? Or, is there another way to model this system so that the solver doesn't need to keep track of the mallet head after the collision?