strange velocity vector
- florante
- Topic Author
- Offline
- Senior Member
Less
More
- Posts: 50
- Thank you received: 0
13 years 7 months ago #4577
by florante
strange velocity vector was created by florante
Hi all,
I'm simulating a simple pipe flow with cs. the result was so strange that there are velocity vectors so large and pointing on a different direction .
Anybody van help on what went wrong with the calculation?
www.flickr.com/photos/37485943@N05/49092...es/s/in/photostream/
Thanks
Florante<br /><br />Post edited by: florante, at: 2010/08/20 10:37
I'm simulating a simple pipe flow with cs. the result was so strange that there are velocity vectors so large and pointing on a different direction .
Anybody van help on what went wrong with the calculation?
www.flickr.com/photos/37485943@N05/49092...es/s/in/photostream/
Thanks
Florante<br /><br />Post edited by: florante, at: 2010/08/20 10:37
Please Log in or Create an account to join the conversation.
- Claus
- Offline
- Moderator
Less
More
- Posts: 670
- Thank you received: 34
13 years 7 months ago #4578
by Claus
Code_Aster release : STA11.4 on OpenSUSE 12.3 64 bits - EDF/Intel version
Replied by Claus on topic Re:strange velocity vector
I think if you look in the 'listing' file you'll find that the solution hasn't converged well. Look for high values under 'derive'.
This can happen for a variety of reasons - how is the setup?
Regards,
Claus
This can happen for a variety of reasons - how is the setup?
Regards,
Claus
Code_Aster release : STA11.4 on OpenSUSE 12.3 64 bits - EDF/Intel version
Please Log in or Create an account to join the conversation.
- florante
- Topic Author
- Offline
- Senior Member
Less
More
- Posts: 50
- Thank you received: 0
13 years 7 months ago #4579
by florante
Replied by florante on topic Re:strange velocity vector
Hi Clause,
Here are two items that are diverging.
[code:1]
Max Values
TurbEner : 0.38100 E+67
Dissip : 0.13786 E+73
[/code:1]
The set up is
Pipe : D=100mm L=200mm
Turbulence mode : k-e
inlet : v=-1m/s T=25oC
Wall : -2 (is this W/m^2 or just watts?)
Reference time step : 0.1s
Mesh salome : Netgen 3D (max size :0.005 fine, second order, optimized)
When I lower the velocity, say to the order of 0.01-0.05 m/s, the calculation goes well. It seems that higher Velocity makes the TurbEner divergent. If this is the case, which parameters do I need to adjust to make TurbEner convergent? or my mesh need more refining or its too fine already ?
Thanks
Flornate
Here are two items that are diverging.
[code:1]
Max Values
TurbEner : 0.38100 E+67
Dissip : 0.13786 E+73
[/code:1]
The set up is
Pipe : D=100mm L=200mm
Turbulence mode : k-e
inlet : v=-1m/s T=25oC
Wall : -2 (is this W/m^2 or just watts?)
Reference time step : 0.1s
Mesh salome : Netgen 3D (max size :0.005 fine, second order, optimized)
When I lower the velocity, say to the order of 0.01-0.05 m/s, the calculation goes well. It seems that higher Velocity makes the TurbEner divergent. If this is the case, which parameters do I need to adjust to make TurbEner convergent? or my mesh need more refining or its too fine already ?
Thanks
Flornate
Please Log in or Create an account to join the conversation.
- Yvan Fournier
- Offline
- Senior Member
Less
More
- Posts: 46
- Thank you received: 6
13 years 7 months ago #4582
by Yvan Fournier
Replied by Yvan Fournier on topic Re:strange velocity vector
Hello,
You may simply be using a time step too large for your mesh. What is the CFL (Courant) number in your case ? The CFD should ideally remain under 5, and a CFL above 20 is definitely risky. If your cell size is in the range of 0.005 meters, with a velocity of 1 m/s, your CFL with a time step of 0.1 s is 1x0.1/0.005 = 20.
You may check th CFL postprocessing with a visualization tool, as the corresponding field is output by default.
Another possible remedy in case of a distorted mesh (often useful for tetrahedral meshes, especially when not all tetrahedra are of good quality) is to add a bit (0.9) of relaxation of pressure increase.
Finally, if you have a tetrahedral mesh with no prism boundary layer, the gradient reconstruction near the wall may not be great, and this may not play well with the turbulence wall model. There are also several wall models to choose from, so testing the sensitivity to those may help, as well as activating/deactivating reconstruction of fluxes at boundaries. All these options should be accessible through the GUI.
Regards,
Yvan
You may simply be using a time step too large for your mesh. What is the CFL (Courant) number in your case ? The CFD should ideally remain under 5, and a CFL above 20 is definitely risky. If your cell size is in the range of 0.005 meters, with a velocity of 1 m/s, your CFL with a time step of 0.1 s is 1x0.1/0.005 = 20.
You may check th CFL postprocessing with a visualization tool, as the corresponding field is output by default.
Another possible remedy in case of a distorted mesh (often useful for tetrahedral meshes, especially when not all tetrahedra are of good quality) is to add a bit (0.9) of relaxation of pressure increase.
Finally, if you have a tetrahedral mesh with no prism boundary layer, the gradient reconstruction near the wall may not be great, and this may not play well with the turbulence wall model. There are also several wall models to choose from, so testing the sensitivity to those may help, as well as activating/deactivating reconstruction of fluxes at boundaries. All these options should be accessible through the GUI.
Regards,
Yvan
Please Log in or Create an account to join the conversation.
- florante
- Topic Author
- Offline
- Senior Member
Less
More
- Posts: 50
- Thank you received: 0
13 years 7 months ago #4589
by florante
Replied by florante on topic Re:strange velocity vector
Thanks Yvan!
I tried to fine tune the time steps now. But because some of the errors occur at 15s in the simulation, I have a hard time verifying if I will not get the same error when I hit the same time frame. But checking the the trend so far, the courant numbers are somewhere in 10-15 max compared to the previous one which are in the range of 65.
thanks
Florante
I tried to fine tune the time steps now. But because some of the errors occur at 15s in the simulation, I have a hard time verifying if I will not get the same error when I hit the same time frame. But checking the the trend so far, the courant numbers are somewhere in 10-15 max compared to the previous one which are in the range of 65.
thanks
Florante
Please Log in or Create an account to join the conversation.
- bean
- Offline
- New Member
Less
More
- Posts: 2
- Thank you received: 0
13 years 1 month ago #5313
by bean
what is seborrheic dermatitis and wig hair , wig at healthy hair high lights
Replied by bean on topic Re:strange velocity vector
Did you succeed step 2?
In that case, please challenge step 3 again.
I rewrote the description in detail.
I am sorry, it is not possible to check it by downloading an individual file
In that case, please challenge step 3 again.
I rewrote the description in detail.
I am sorry, it is not possible to check it by downloading an individual file
what is seborrheic dermatitis and wig hair , wig at healthy hair high lights
Please Log in or Create an account to join the conversation.
Moderators: catux
Time to create page: 0.174 seconds