Time-history analisys
11 years 1 month ago - 11 years 1 month ago #7492
by Jay Cloud
Replied by Jay Cloud on topic Re: Time-history analisys
Hello again,
I posted in Code-Aster forums as you told me. It looks like there is a bug with the version of ASTK/Code-Aster I was using or something like that. I didn't know exactly what to do to solve it so I installed the new CAELinux 2013 and problem solved.
Thank you very much for all your help again.
So with this new version I'm trying to run the same analysis. And this is the new problem I have:
This message appears in the .mess file a few times (same error with maille M2, maille M3 and so on, and it stops after printing it five times saying something like no more outputs for the same error).
I think it is something with the top free border of the cylinder, but I don't completely understand what it means so I don't know how to solve it (the same question is posted here but no solution was given since the author asked another thing).
However the analysis is partly done. After running it with ASTK, a table.txt file (see atachment) with node displacement is created. But I can't use the Post-Pro Module in Salome to see the deformed shape.
Since I'm using a newer version of Code-Aster, I modified the provided .comm file (also attached) in the Wiki website. I "copied" the provided one and just changed a few functions not recognized by Eficas (for example, MACRO_MATR_ASSE replaced by ASSEMBLAGE if I am not mistaken). Therefore I think there is no problem with the syntax of the .comm file, but just in case.
And another two "problems" I had (not really important, but just curious):
- I can't use "Create Group from Geometry" in the Mesh Module to create groups of "Surfin" and "Ctop" (it works with "Cbot" though). Is there any explanation (.hdf & .py files attached)? I had to create those groups since I got an error related to that (I guess, after creating them I did not have that error again).
- With this version of Eficas, I can't create a new parameter using another two parameters to define its numerical value. That is, if I create 'Param_A=1', 'Param_B=2', I can't create 'Param_C=Param_A+Param_B'. Eficas won't let me, I get 'valeur incorrecte' error. I have to set 'Param_C=3' instead.
Thank you.
Regards,
Jay.
EDIT: I can't attach any file. ???? Is it allowed to provide a download link (zippyshare, mediafire...)??
I posted in Code-Aster forums as you told me. It looks like there is a bug with the version of ASTK/Code-Aster I was using or something like that. I didn't know exactly what to do to solve it so I installed the new CAELinux 2013 and problem solved.
Thank you very much for all your help again.
So with this new version I'm trying to run the same analysis. And this is the new problem I have:
!-------------------------------------------------------------------------------------!
! <A> <CALCULEL2_63> !
! !
! -> La maille M1 porte un élément fini de bord, mais elle ne borde !
! aucun élément ayant une "rigidité". !
! !
! -> Risque & Conseil : !
! Cela peut entraîner des problèmes de "pivot nul" lors de la résolution. !
! Si la résolution des systèmes linéaires ne pose pas de problèmes, vous !
! pouvez ignorer ce message. !
! Sinon, vérifier la définition du modèle (AFFE_MODELE) en évitant l'utilisation !
! de l'opérande TOUT='OUI'. !
! !
! !
! Ceci est une alarme. Si vous ne comprenez pas le sens de cette !
! alarme, vous pouvez obtenir des résultats inattendus ! !
!-------------------------------------------------------------------------------------!I think it is something with the top free border of the cylinder, but I don't completely understand what it means so I don't know how to solve it (the same question is posted here but no solution was given since the author asked another thing).
However the analysis is partly done. After running it with ASTK, a table.txt file (see atachment) with node displacement is created. But I can't use the Post-Pro Module in Salome to see the deformed shape.
Since I'm using a newer version of Code-Aster, I modified the provided .comm file (also attached) in the Wiki website. I "copied" the provided one and just changed a few functions not recognized by Eficas (for example, MACRO_MATR_ASSE replaced by ASSEMBLAGE if I am not mistaken). Therefore I think there is no problem with the syntax of the .comm file, but just in case.
And another two "problems" I had (not really important, but just curious):
- I can't use "Create Group from Geometry" in the Mesh Module to create groups of "Surfin" and "Ctop" (it works with "Cbot" though). Is there any explanation (.hdf & .py files attached)? I had to create those groups since I got an error related to that (I guess, after creating them I did not have that error again).
- With this version of Eficas, I can't create a new parameter using another two parameters to define its numerical value. That is, if I create 'Param_A=1', 'Param_B=2', I can't create 'Param_C=Param_A+Param_B'. Eficas won't let me, I get 'valeur incorrecte' error. I have to set 'Param_C=3' instead.
Thank you.
Regards,
Jay.
EDIT: I can't attach any file. ???? Is it allowed to provide a download link (zippyshare, mediafire...)??
11 years 1 month ago #7494
by Claus
Code_Aster release : STA11.4 on OpenSUSE 12.3 64 bits - EDF/Intel version
Replied by Claus on topic Re: Time-history analisys
Okay, great 
I think the first error means that an element doesn't have a element type assigned to it, or something like that. It's hard to guess without the files.
Why can't you use the post-pro to view the deformed shape? what's the error?
Regarding EFICAS: yes well, this program has these errors from time to time and version to version. I can't really tell you what causes your problem, but try saving the file, add the parameters with a text editor and reopening the file. Not ideal, but it should work.
You can link to external download sites, but it is much, much more preferable if you can figure out how to reduce the size of the files. Otherwise the links will be dead sometime in the future when someone has a similar problem
Does the .py file generate the study? if so, just attach that.
Is the mesh saved with the HDF? if so, clear the mesh before saving the HDF and tell us how to generate it and recreate the problem.
/C
I think the first error means that an element doesn't have a element type assigned to it, or something like that. It's hard to guess without the files.
Why can't you use the post-pro to view the deformed shape? what's the error?
Regarding EFICAS: yes well, this program has these errors from time to time and version to version. I can't really tell you what causes your problem, but try saving the file, add the parameters with a text editor and reopening the file. Not ideal, but it should work.
You can link to external download sites, but it is much, much more preferable if you can figure out how to reduce the size of the files. Otherwise the links will be dead sometime in the future when someone has a similar problem
Does the .py file generate the study? if so, just attach that.
Is the mesh saved with the HDF? if so, clear the mesh before saving the HDF and tell us how to generate it and recreate the problem.
/C
Code_Aster release : STA11.4 on OpenSUSE 12.3 64 bits - EDF/Intel version
11 years 1 month ago - 11 years 1 month ago #7495
by Jay Cloud
Replied by Jay Cloud on topic Re: Time-history analisys
Hello,
Yes, hard to guess without the files. But I couldn't attach them. I don't think it was because of the size, it was 50 kb or so. I even tried to attach just one file and I could not either. =/
Actually, the files are the same as those provided in the Wiki page ( here ). Maybe now you use them and you can use "Create Groups from Geometry" without any problem...
I have a question related to this. In order to write a .comm file, the 'Groups' used to apply forces, impose degree of freedom and such functions, have to be defined in both Geometry (by 'Explode' or 'Create group') and Mesh ('Create group (from geometry or not)')? Only in one of them (which one)?
I thought that having those groups defined in the Geometry was enough, but as I said in my last post, I had en error due to a 'non-existing' mesh (which was defined in the Geometry but not in the Mesh).
I couldn't use the Post-Pro because its icon didn't appear in the object browser after running the analysis with ASTK.
* My fault. I'm not sure wether after running ASTK a .resu.med file was created or not, but I didn't know I could open it directly from the Post-Pro module in Salome... But, if it did not appear in the object browser, I guess it was due to an error.
And now a new case. The actual analysis I have to run for my experiment. I managed to successfully run it, so it's something.
But I have new questions as always.
I "copied" the .comm file. But since my geometry is different I had to remove some functions.
- Do I have to use the function CREA_MAILLAGE / ORIE_PEAU_XX anywhere in my geometry?
- Do I have to add the function AFFE_CARA_ELEM for the prism? I read the documentation U4.42.01 , maybe I need to use 'POUTRE', 'RECTANGLE' for the prism...)
- When using AFFE_CHAR_MECA, is it correct to use PRES_REP for the applied force, together with DEFI_FONCTION? I mean, I am not sure if PRES_REP represents a constant (in time) pressure, but if I use it together with DEFI_FONCTION, is it possible to define a function (similar to a Heaviside Step function)? With FORCE_ARETE all the displacements in the output file were zero (the force is applied in a surface, not an edge, so I guess FORCE_ARETE is not the correct function).
As I said, I run the analysis and it works. But maybe adding any of those functions (or even others) I can get 'better' results.
In the attached file (when I can upload it) are the .hdf, .astk & .comm file used for the new geometry. There is also a Time vs Z-displacement plot screenshot of a node in the prism. It is not what I expected, I expected some more "regular" displacements (with Max(DZ) approximately the same as Abs(Min(DZ)) for every period).
Thank you.
Regards,
Jay
PS: I think it is a problem with the web browser. I'll try to attach the files later using another computer.
EDIT: Nope. I still can't attach files. I've uploaded it to ZippyShare . I'll try again another day and I'll attach it if I can.
Yes, hard to guess without the files. But I couldn't attach them. I don't think it was because of the size, it was 50 kb or so. I even tried to attach just one file and I could not either. =/
Actually, the files are the same as those provided in the Wiki page ( here ). Maybe now you use them and you can use "Create Groups from Geometry" without any problem...
I have a question related to this. In order to write a .comm file, the 'Groups' used to apply forces, impose degree of freedom and such functions, have to be defined in both Geometry (by 'Explode' or 'Create group') and Mesh ('Create group (from geometry or not)')? Only in one of them (which one)?
I thought that having those groups defined in the Geometry was enough, but as I said in my last post, I had en error due to a 'non-existing' mesh (which was defined in the Geometry but not in the Mesh).
I couldn't use the Post-Pro because its icon didn't appear in the object browser after running the analysis with ASTK.
* My fault. I'm not sure wether after running ASTK a .resu.med file was created or not, but I didn't know I could open it directly from the Post-Pro module in Salome... But, if it did not appear in the object browser, I guess it was due to an error.
And now a new case. The actual analysis I have to run for my experiment. I managed to successfully run it, so it's something.
But I have new questions as always.
I "copied" the .comm file. But since my geometry is different I had to remove some functions.
- Do I have to use the function CREA_MAILLAGE / ORIE_PEAU_XX anywhere in my geometry?
- Do I have to add the function AFFE_CARA_ELEM for the prism? I read the documentation U4.42.01 , maybe I need to use 'POUTRE', 'RECTANGLE' for the prism...)
- When using AFFE_CHAR_MECA, is it correct to use PRES_REP for the applied force, together with DEFI_FONCTION? I mean, I am not sure if PRES_REP represents a constant (in time) pressure, but if I use it together with DEFI_FONCTION, is it possible to define a function (similar to a Heaviside Step function)? With FORCE_ARETE all the displacements in the output file were zero (the force is applied in a surface, not an edge, so I guess FORCE_ARETE is not the correct function).
As I said, I run the analysis and it works. But maybe adding any of those functions (or even others) I can get 'better' results.
In the attached file (when I can upload it) are the .hdf, .astk & .comm file used for the new geometry. There is also a Time vs Z-displacement plot screenshot of a node in the prism. It is not what I expected, I expected some more "regular" displacements (with Max(DZ) approximately the same as Abs(Min(DZ)) for every period).
Thank you.
Regards,
Jay
PS: I think it is a problem with the web browser. I'll try to attach the files later using another computer.
EDIT: Nope. I still can't attach files. I've uploaded it to ZippyShare . I'll try again another day and I'll attach it if I can.
11 years 1 month ago #7496
by Claus
Code_Aster release : STA11.4 on OpenSUSE 12.3 64 bits - EDF/Intel version
Replied by Claus on topic Re: Time-history analisys
Lots of questions - I'll try to take them one by one
I just tried the files from the wiki, and I have the same problem. The groups are empty judging from the python error.
Just re-create your own geometry, since it's so simple.
Regarding groups: If you use SaloméMECA wizards to create command files, you can use the geometrical groups only - the wizard updates the mesh with the groups before it exports it.
Any other time you must create mesh groups. Code_Aster is not aware that the mesh is created with MECA and has no knowledge of the HDF file, only the MED (or unv, msh, etc.) file.
The result file will only appear in MECA if the computation ran successfully, but at any time you can open the post-pro module and import any file.
The CREA_MAILLAGE / ORIE_PEAU_XX command reorients surfaces to have the same 'normals' in case the mesher (in this case MECA) didn't do a proper job. Sometimes it's needed, sometimes it's not. Code_Aster will usually throw an error if it detects invalid surfaces.
Im not sure which files / test case you are trying to copy, so I don't know if you need beam elements etc.
PRES_REP applies a pressure to surfaces, so you need to know your surface area, if you want to talk about (resultant) forces.
If you are running a nonlinear simulation (dynamic or quasi static) with pressure as a function of time, you must apply it gradually. These solvers do not (as far as I know) support step functions. If you want to calculate a response from such an applied force, you must resort to a transient calculation, but that's not something I know too much about yet.
FORCE_ARETE is correctly online for line segments. Im surprised the simulation ran..
Lets see if that answered some of your questions
Edit: I just saw your uploaded files, and indeed you're using DYNA_LINE_TRAN, so now you can tell me how to apply a stepfunction
I just tried the files from the wiki, and I have the same problem. The groups are empty judging from the python error.
Just re-create your own geometry, since it's so simple.
Regarding groups: If you use SaloméMECA wizards to create command files, you can use the geometrical groups only - the wizard updates the mesh with the groups before it exports it.
Any other time you must create mesh groups. Code_Aster is not aware that the mesh is created with MECA and has no knowledge of the HDF file, only the MED (or unv, msh, etc.) file.
The result file will only appear in MECA if the computation ran successfully, but at any time you can open the post-pro module and import any file.
The CREA_MAILLAGE / ORIE_PEAU_XX command reorients surfaces to have the same 'normals' in case the mesher (in this case MECA) didn't do a proper job. Sometimes it's needed, sometimes it's not. Code_Aster will usually throw an error if it detects invalid surfaces.
Im not sure which files / test case you are trying to copy, so I don't know if you need beam elements etc.
PRES_REP applies a pressure to surfaces, so you need to know your surface area, if you want to talk about (resultant) forces.
If you are running a nonlinear simulation (dynamic or quasi static) with pressure as a function of time, you must apply it gradually. These solvers do not (as far as I know) support step functions. If you want to calculate a response from such an applied force, you must resort to a transient calculation, but that's not something I know too much about yet.
FORCE_ARETE is correctly online for line segments. Im surprised the simulation ran..
Lets see if that answered some of your questions
Edit: I just saw your uploaded files, and indeed you're using DYNA_LINE_TRAN, so now you can tell me how to apply a stepfunction
Code_Aster release : STA11.4 on OpenSUSE 12.3 64 bits - EDF/Intel version
10 years 11 months ago #7580
by Jay Cloud
Replied by Jay Cloud on topic Re: Time-history analisys
Hello,
It's been a month since my last post here, but I think it is better to continue using this thread instead of starting a new one.
From the last post:
So, I am with the time-history analysis. With PRINT_RESU I get displacements (DX, DY, DZ) of a node to see the reaction after an impact. Plotting displacement VS time I expected something like this . However, I guess the number of DOF in my "structure" is bigger than in the structure of the picture .
According to the COMM file, the applied force is FZ = -xxx N (negative). Then applying the "step-function" with DEFI_LIST_REEL & DEFI_FONCTION, the force FZ is applied from t = t1 to t = te2.
If I plot DXYZ vs Time with time from 0.0 to te = 0.04, that is, the duration of the impact, this is what I get:
Honestly, I don't know if those displacements can be considered as normal.
But, if I plot from t = 0.0 to t = 0.1, so I can see the displacements of a node after the impact (something like a "free-vibration"), I get this:
There is a big difference between displacements during and after the impact.
And another question, how do I get the values for AMOR_ALPHA & AMOR_BETA (damping coefficients)? I tried with this and this , but nothing.
I cannot attach anything, so: HDF, COMM & ASTK files (there is also a file with the first 20 frequencies from a modal analysis).
Thanks in advance. Regards,
Jay
It's been a month since my last post here, but I think it is better to continue using this thread instead of starting a new one.
From the last post:
I'll try to explain it later in this post, since I'm not sure if what I am doing is correct.claws wrote: Edit: I just saw your uploaded files, and indeed you're using DYNA_LINE_TRAN, so now you can tell me how to apply a stepfunction
So, I am with the time-history analysis. With PRINT_RESU I get displacements (DX, DY, DZ) of a node to see the reaction after an impact. Plotting displacement VS time I expected something like this . However, I guess the number of DOF in my "structure" is bigger than in the structure of the picture .
According to the COMM file, the applied force is FZ = -xxx N (negative). Then applying the "step-function" with DEFI_LIST_REEL & DEFI_FONCTION, the force FZ is applied from t = t1 to t = te2.
tsteps = 500; to = 0.0; tc = 0.002; t1 = 0.00200002;
te = 0.04; te2 = 0.040004; tf = 0.1
time=DEFI_LIST_REEL(DEBUT=to,
INTERVALLE=_F(JUSQU_A=tf,
NOMBRE=tsteps,),
INFO=1,
TITRE='time',);
castle=DEFI_FONCTION(
NOM_PARA='INST',
VALE=(to,0.0, # this is a "multiplication factor"
tc,0.0, # so, from 0.0 to 0.002 seconds, the applied force is FZ * 0 = 0
t1,1.00, # from 0.00200002 to
te,1.00, # 0.04 the applied force is FZ * 1 = FZ
te2,0.00, # and from 0.040004 onwards, FZ * 0 = 0
tf,0.00,
),
INFO=1,
TITRE='castle',);If I plot DXYZ vs Time with time from 0.0 to te = 0.04, that is, the duration of the impact, this is what I get:
Honestly, I don't know if those displacements can be considered as normal.
But, if I plot from t = 0.0 to t = 0.1, so I can see the displacements of a node after the impact (something like a "free-vibration"), I get this:
There is a big difference between displacements during and after the impact.
And another question, how do I get the values for AMOR_ALPHA & AMOR_BETA (damping coefficients)? I tried with this and this , but nothing.
I cannot attach anything, so: HDF, COMM & ASTK files (there is also a file with the first 20 frequencies from a modal analysis).
Thanks in advance. Regards,
Jay
Moderators: catux
Time to create page: 0.143 seconds