Problem of MACR_RECAL
- Samuel
- Topic Author
- Offline
- New Member
Less
More
- Posts: 4
- Thank you received: 0
9 years 9 months ago #7836
by Samuel
Problem of MACR_RECAL was created by Samuel
Hello everyone! I have a problem about the usage of MACR_RECAL.
I use it to iterate a slave program of calculating the P-t curve of indentation for approaching the experiment curve which is also calculated by the slave program. For verification of the uniqueness of 6 parameters in the simplified Chaboche model, I try to use the exact parameters for the initial values. But there isn't any convergence. The Residual of the Iteration 0 equal to 1.0 .
I don't understand why the residual of Iteration 0 equal to 1 because it is exactly the same curve... How does it calculate the residual?
I really need your help! Thank you!
I use it to iterate a slave program of calculating the P-t curve of indentation for approaching the experiment curve which is also calculated by the slave program. For verification of the uniqueness of 6 parameters in the simplified Chaboche model, I try to use the exact parameters for the initial values. But there isn't any convergence. The Residual of the Iteration 0 equal to 1.0 .
I don't understand why the residual of Iteration 0 equal to 1 because it is exactly the same curve... How does it calculate the residual?
I really need your help! Thank you!
- RichardS
- Offline
- Elite Member
9 years 9 months ago #7839
by RichardS
SimScale - Engineering Simulation in your browser!
Replied by RichardS on topic Re: Problem of MACR_RECAL
Hello,
without any files it is very hard to tell what's the problem.
You should provide the files to re-run the simulation or at least the message file of the failing run.
Best regards,
Richard
without any files it is very hard to tell what's the problem.
You should provide the files to re-run the simulation or at least the message file of the failing run.
Best regards,
Richard
SimScale - Engineering Simulation in your browser!
The following user(s) said Thank You: Samuel
- Samuel
- Topic Author
- Offline
- New Member
Less
More
- Posts: 4
- Thank you received: 0
9 years 9 months ago - 9 years 9 months ago #7840
by Samuel
Replied by Samuel on topic Re: Problem of MACR_RECAL
Oh! I'm sorry, I forgot to give the files!
Here are the main program, slave program and the mesh I used.
sphericalindent_chaboche_force_main.comm
sphericalindent_chaboche_force_slave.3.comm
sphericalindent_chaboche_force.med
I use the slave program to draw a P-t curve according to the parameters below: (assumed to be the experiment data)
ETA : 140.14
EXP_N : 16.5436023657
GAMMA : 15796.0723344
PETITB : 400.0
R0 : 123.420729958
YOUNG : 116139.140085
And then I run the main program with the initial values which is the exact values before.
Below there is the convergence information from .mess
Informations de convergence :
======================================================================
=> Norme de l'erreur = 4.2822276827e-11
Informations de convergence :
======================================================================
======================================================================
Iteration 0 :
=> Fonctionnelle = 1.0
=> Residu = 1.0
=> Paramètres :
ETA : 140.14
EXP_N : 16.5436023657
GAMMA : 15796.0723344
PETITB : 400.0
R0 : 123.420729958
YOUNG : 116139.140085
======================================================================
Informations de convergence :
======================================================================
=> Norme de l'erreur = 3.36030287054e-11
Informations de convergence :
======================================================================
======================================================================
Iteration 100 :
=> Fonctionnelle = 0.784709062556
=> Residu = 0.0255398575846
=> Variation des parametres (norme L2) = 0.0657683553719
=> Variation de la fonctionnelle = 0.000663919241281
=> Paramètres :
ETA : 140.140883294
EXP_N : 16.5436421652
GAMMA : 15797.6446903
PETITB : 400.031133701
R0 : 123.419244519
YOUNG : 116139.136337
======================================================================
======================================================================
CONVERGENCE NON ATTEINTE
--> Le nombre maximum d'iterations de l'algorithme (ITER_MAXI) a ete atteint <--
======================================================================
We can see that the error of the norm is small enough at the beginning, (in fact they are exactly the same curve for the iteration 0!) but the residual we got is 1, which seems to accord with the statistics. But in the main program, we require residual<0.0001, which is a default. We can not control the error of the norm in the main program which seems more reasonable to approch the experiment curve.
So I want to know why we can not recognize the exact values with the MACR_RECAL?
Thank you for your help!
Best wishes!
Here are the main program, slave program and the mesh I used.
sphericalindent_chaboche_force_main.comm
sphericalindent_chaboche_force_slave.3.comm
sphericalindent_chaboche_force.med
I use the slave program to draw a P-t curve according to the parameters below: (assumed to be the experiment data)
ETA : 140.14
EXP_N : 16.5436023657
GAMMA : 15796.0723344
PETITB : 400.0
R0 : 123.420729958
YOUNG : 116139.140085
And then I run the main program with the initial values which is the exact values before.
Below there is the convergence information from .mess
Informations de convergence :
======================================================================
=> Norme de l'erreur = 4.2822276827e-11
Informations de convergence :
======================================================================
======================================================================
Iteration 0 :
=> Fonctionnelle = 1.0
=> Residu = 1.0
=> Paramètres :
ETA : 140.14
EXP_N : 16.5436023657
GAMMA : 15796.0723344
PETITB : 400.0
R0 : 123.420729958
YOUNG : 116139.140085
======================================================================
Informations de convergence :
======================================================================
=> Norme de l'erreur = 3.36030287054e-11
Informations de convergence :
======================================================================
======================================================================
Iteration 100 :
=> Fonctionnelle = 0.784709062556
=> Residu = 0.0255398575846
=> Variation des parametres (norme L2) = 0.0657683553719
=> Variation de la fonctionnelle = 0.000663919241281
=> Paramètres :
ETA : 140.140883294
EXP_N : 16.5436421652
GAMMA : 15797.6446903
PETITB : 400.031133701
R0 : 123.419244519
YOUNG : 116139.136337
======================================================================
======================================================================
CONVERGENCE NON ATTEINTE
--> Le nombre maximum d'iterations de l'algorithme (ITER_MAXI) a ete atteint <--
======================================================================
We can see that the error of the norm is small enough at the beginning, (in fact they are exactly the same curve for the iteration 0!) but the residual we got is 1, which seems to accord with the statistics. But in the main program, we require residual<0.0001, which is a default. We can not control the error of the norm in the main program which seems more reasonable to approch the experiment curve.
So I want to know why we can not recognize the exact values with the MACR_RECAL?
Thank you for your help!
Best wishes!
Last edit: 9 years 9 months ago by Samuel. Reason: Files missing.
- Samuel
- Topic Author
- Offline
- New Member
Less
More
- Posts: 4
- Thank you received: 0
9 years 9 months ago #7841
by Samuel
Replied by Samuel on topic Re: Problem of MACR_RECAL
sphericalindent_chaboche_force_main.comm
sphericalindent_chaboche_force_slave.3.comm
sphericalindent_chaboche_force.med
sphericalindent_chaboche_force_slave.3.comm
sphericalindent_chaboche_force.med
- Samuel
- Topic Author
- Offline
- New Member
Less
More
- Posts: 4
- Thank you received: 0
9 years 9 months ago #7842
by Samuel
Replied by Samuel on topic Re: Problem of MACR_RECAL
I don't know why I can't upload the files, if you want the program, I can send you by e-mail.
Moderators: catux
Time to create page: 0.155 seconds