- Forum
- Public Forums
- Finite Element Modelling - Code_Aster
- Dynamic analysis of an axial loaded cylinder
Dynamic analysis of an axial loaded cylinder
- JMB
- Topic Author
- Offline
- Elite Member
-
Less
More
- Posts: 166
- Thank you received: 0
15 years 8 months ago #3482
by JMB
Dynamic analysis of an axial loaded cylinder was created by JMB
Hello Kees Wouters,
Thank you for the well illustrated and detailed tutorials posted at:
www.caelinux.org/wiki/index.php/Contrib:KeesWouters
Browsing through the one at: www.caelinux.org/wiki/index.php/Contrib:KeesWouters/dynamicload
it seems to me that ending time stated (300 ms) is not consistent with the CodeAster commands described later in the tutorial. Should it not be 20 ms, because te=0.02? Also according to the commands the load will be zero in the time interval 0 to 0.0050004999999...), should it not? You have stated the load is zero in the interval 0 - 100 ms. I have reproduced your tutorial below in this email for your review, comment or correction.
Regards,
JMB
=====================================================================
The load is an axial force on the top edge of the cylinder. A scaling factor is applied by the MULT_FONCT command in the command DYNA_LINE_TRAN. This scaling factor is zero for the first 100 ms (How?), and goes to unity within a very small time scale, ie. small compared to the resonance frequencies of the axial modes of the cylinder. The scale factor remains unity until the end of the calculation at 300 ms (How?).
A time step is defined by the DEFI_LIST_REEL command. The scaling function castle is defined by the DEFI_FONCTION command. In total tsteps (200) output times will be generated.
# The load will be applied in steps defined by a 'time' function and
# multiplication factor on the load.
tsteps = 200
to = 0.000
tc = 0.005
t1 = (1+1e-5)*tc
te = 0.020
time=DEFI_LIST_REEL(DEBUT=0.0,
INTERVALLE=_F(JUSQU_A=te,NOMBRE=tsteps,),
INFO=2,TITRE='time',);
castle=DEFI_FONCTION(NOM_PARA='INST',
VALE=(to, 0.00,
tc, 0.00,
t1 , 1.00,
te, 1.00,),
INFO=2,TITRE='castle',);
Thank you for the well illustrated and detailed tutorials posted at:
www.caelinux.org/wiki/index.php/Contrib:KeesWouters
Browsing through the one at: www.caelinux.org/wiki/index.php/Contrib:KeesWouters/dynamicload
it seems to me that ending time stated (300 ms) is not consistent with the CodeAster commands described later in the tutorial. Should it not be 20 ms, because te=0.02? Also according to the commands the load will be zero in the time interval 0 to 0.0050004999999...), should it not? You have stated the load is zero in the interval 0 - 100 ms. I have reproduced your tutorial below in this email for your review, comment or correction.
Regards,
JMB
=====================================================================
The load is an axial force on the top edge of the cylinder. A scaling factor is applied by the MULT_FONCT command in the command DYNA_LINE_TRAN. This scaling factor is zero for the first 100 ms (How?), and goes to unity within a very small time scale, ie. small compared to the resonance frequencies of the axial modes of the cylinder. The scale factor remains unity until the end of the calculation at 300 ms (How?).
A time step is defined by the DEFI_LIST_REEL command. The scaling function castle is defined by the DEFI_FONCTION command. In total tsteps (200) output times will be generated.
# The load will be applied in steps defined by a 'time' function and
# multiplication factor on the load.
tsteps = 200
to = 0.000
tc = 0.005
t1 = (1+1e-5)*tc
te = 0.020
time=DEFI_LIST_REEL(DEBUT=0.0,
INTERVALLE=_F(JUSQU_A=te,NOMBRE=tsteps,),
INFO=2,TITRE='time',);
castle=DEFI_FONCTION(NOM_PARA='INST',
VALE=(to, 0.00,
tc, 0.00,
t1 , 1.00,
te, 1.00,),
INFO=2,TITRE='castle',);
- kwou
-
- Offline
- Moderator
-
15 years 8 months ago #3483
by kwou
Interest: structural mechanics, solar energy (picture at 'my location' shows too little pv panels)
--
kind regards - kees
Replied by kwou on topic Re:Dynamic analysis of an axial loaded cylinder
Hoi JMB,
Your are quite right. The correct timings for the given picture are:
tsteps = 400
to = 0.000
tc = 0.002
t1 = (1+1e-5)*tc
te = 0.040
time=DEFI_LIST_REEL(DEBUT=to,
INTERVALLE=_F(JUSQU_A=te,NOMBRE=tsteps,),
INFO=2,TITRE='time',);
castle=DEFI_FONCTION(NOM_PARA='INST',
VALE=(to,0.00,tc,0.00,t1,1.00,te,1.00,),
INFO=2,TITRE='castle',);
(I hope no mistakes this time; updated command and text files should be there now.)
To summarise (from ...VALE=(to,0.00,tc,0.00,t1,1.00,te,1.00,)...):
- no load between 0 and 2 ms (from to to tc)
- ramping between 2 and 2+ ms (from tc to t1)
- unit load between 2+ and 40 ms (from t1 to te)
I changed the end time during the preparation of the tutorial from 20 to 40 ms because the first five cycles indicated a decreasing amplitude. However, after five cycles the amplitude remains more or less the same. That is why I changed the timings.
Furthermore, I ask the question about the damping matrix there: what damping 'coefficients' are used for this matrix? I tried to read and understand the explanation given in R5.05.04/U2.06.03/U4.61.01/U4.43.01 but I couldnot really figure out the values for AMOR_ALPHA/BETA.
This is from U4.43.01, but no default values are given for AMOR_ALPHA/BETA ..., and I didnot give them any values in the command files. So are these values set to 0?
3.1.1 Syntaxe
|/ ELAS = _F ( ♦ E = yg, [R]
♦ NU = nu, [R]
◊ RHO = rho, [R]
◊ ALPHA = dil , [R]
◊ AMOR_ALPHA = a_alpha, [R]
◊ AMOR_BETA = a_beta, [R]
◊ AMOR_HYST = eta [R]
Added saterday:
indeed, if no values for parameters AMOR-ALPHA/BETA are given, they are zero and the damping matrix is zero, ie no damping applied. To add (viscous) damping to the system, in the material definition put something like:
a_alpha = 2.5e-4
a_beta = 1000.0
steel=DEFI_MATERIAU(ELAS=_F(E=2.100e5,NU=0.28,RHO=0.007850,AMOR_ALPHA=a_alpha,AMOR_BETA=a_beta,),);
where a_alpha/beta are coefficients for damping proportional to stiffness and mass matrices, again see reference R5.05.04.
kind regards - kees<br /><br />Post edited by: Kees Wouters, at: 2009/10/03 22:17
Your are quite right. The correct timings for the given picture are:
tsteps = 400
to = 0.000
tc = 0.002
t1 = (1+1e-5)*tc
te = 0.040
time=DEFI_LIST_REEL(DEBUT=to,
INTERVALLE=_F(JUSQU_A=te,NOMBRE=tsteps,),
INFO=2,TITRE='time',);
castle=DEFI_FONCTION(NOM_PARA='INST',
VALE=(to,0.00,tc,0.00,t1,1.00,te,1.00,),
INFO=2,TITRE='castle',);
(I hope no mistakes this time; updated command and text files should be there now.)
To summarise (from ...VALE=(to,0.00,tc,0.00,t1,1.00,te,1.00,)...):
- no load between 0 and 2 ms (from to to tc)
- ramping between 2 and 2+ ms (from tc to t1)
- unit load between 2+ and 40 ms (from t1 to te)
I changed the end time during the preparation of the tutorial from 20 to 40 ms because the first five cycles indicated a decreasing amplitude. However, after five cycles the amplitude remains more or less the same. That is why I changed the timings.
Furthermore, I ask the question about the damping matrix there: what damping 'coefficients' are used for this matrix? I tried to read and understand the explanation given in R5.05.04/U2.06.03/U4.61.01/U4.43.01 but I couldnot really figure out the values for AMOR_ALPHA/BETA.
This is from U4.43.01, but no default values are given for AMOR_ALPHA/BETA ..., and I didnot give them any values in the command files. So are these values set to 0?
3.1.1 Syntaxe
|/ ELAS = _F ( &#9830; E = yg, [R]
&#9830; NU = nu, [R]
&#9674; RHO = rho, [R]
&#9674; ALPHA = dil , [R]
&#9674; AMOR_ALPHA = a_alpha, [R]
&#9674; AMOR_BETA = a_beta, [R]
&#9674; AMOR_HYST = eta [R]
Added saterday:
indeed, if no values for parameters AMOR-ALPHA/BETA are given, they are zero and the damping matrix is zero, ie no damping applied. To add (viscous) damping to the system, in the material definition put something like:
a_alpha = 2.5e-4
a_beta = 1000.0
steel=DEFI_MATERIAU(ELAS=_F(E=2.100e5,NU=0.28,RHO=0.007850,AMOR_ALPHA=a_alpha,AMOR_BETA=a_beta,),);
where a_alpha/beta are coefficients for damping proportional to stiffness and mass matrices, again see reference R5.05.04.
kind regards - kees<br /><br />Post edited by: Kees Wouters, at: 2009/10/03 22:17
Interest: structural mechanics, solar energy (picture at 'my location' shows too little pv panels)
--
kind regards - kees
Moderators: catux
- Forum
- Public Forums
- Finite Element Modelling - Code_Aster
- Dynamic analysis of an axial loaded cylinder
Time to create page: 0.155 seconds