[mohsen.samie]

hi. im trying to simulate a transient submerged hydraulic jump but im not sure about the transient convergence i sent some pics of residual convergence of my simulate. is it good or not? caz its different with steady state and i cant find any criteria for transient convergence.
best regards
http://www.4shared.com/photo/TYAEq5y9ce/1_online.html
http://www.4shared.com/photo/uc_vuYwoba/2_online.html
http://www.4shared.com/photo/hofr3NfNce/3_online.html
http://www.4shared.com/photo/U_UPtUloba/4_online.html
http://www.4shared.com/photo/1kYygJ3Ace/5_online.html

[Antanas]

Originally Posted by mohsen.samie

hi. im trying to simulate a transient submerged hydraulic jump but im not sure about the transient convergence i sent some pics of residual convergence of my simulate. is it good or not? caz its different with steady state and i cant find any criteria for transient convergence.
best regards
http://www.4shared.com/photo/TYAEq5y9ce/1_online.html
http://www.4shared.com/photo/uc_vuYwoba/2_online.html
http://www.4shared.com/photo/hofr3NfNce/3_online.html
http://www.4shared.com/photo/U_UPtUloba/4_online.html
http://www.4shared.com/photo/1kYygJ3Ace/5_online.html

Criteria are the same as in steady state. Monitor residuals, imbalances and quantities of interest. Residuals and imbalances must fall below target values at the end of each timestep. That means that if coefficient loops convergence monitoring is not activated then residuals curves must lie below target value.

[pkladisios]

What is your residual target (found in CFX-pre, solver control)? If it's default (1.0e-04), it looks like it's fluctuating around your solution. Also, as Antanas mentioned imbalances are importand. An additional image would help. While solver manager is running, right-click next to the tabs, add new monitor and create an imbalances plot.

[mohsen.samie]

Originally Posted by Antanas

Criteria are the same as in steady state. Monitor residuals, imbalances and quantities of interest. Residuals and imbalances must fall below target values at the end of each timestep. That means that if coefficient loops convergence monitoring is not activated then residuals curves must lie below target value.

tnx dear Antanas. in each time step its falling down but entire of residuals its not good. right? the target value is default 1.0e-4
the RSM courant number is about 0.67 and max courant number is about 10.2 in each time step. wich one of them is criteria between 2 and 10 as i know...max courant number right? can i make the convergence better with better courant number? how much is good for RSM and MAX courant number?

[mohsen.samie]

Originally Posted by pkladisios

What is your residual target (found in CFX-pre, solver control)? If it's default (1.0e-04), it looks like it's fluctuating around your solution. Also, as Antanas mentioned imbalances are importand. An additional image would help. While solver manager is running, right-click next to the tabs, add new monitor and create an imbalances plot.

tnx dear pkladisios.yes its default 1.0e-4
in new monitor there is just plot monitor - residual monitor - text monitor and time corrected monitor....what do u mean by imbalances monitor with making new monitor?
can i improve the convergence with reducing timestep?
how much is good for RSM and MAX courant number?

[Antanas]

Originally Posted by mohsen.samie

in each time step its falling down but entire of residuals its not good. right?

Why? You set target = 1e-4 and residuals are below it at the end of timesteps. It's normal that residuals jump up at the beginning of each timstep. They must decrease during coefficient loops.
Note that for final results it's recommended to set RMS residuals target to <= 1e-5 or MAX residuals <= 1e-4. If target values can not be reached within 3-5 coef. loops it's recommended to reduce timestep size rather than increase number of loops. However, complex physics may require more loops.

Originally Posted by mohsen.samie

how much is good for RSM and MAX courant number?

Recommended average values for CN are 2 – 10, in some cases higher values are acceptable.

[pkladisios]

From my limited, mind you, understanding i think that residuals must progressively drop below or move along your residual target. In your case, the residuals move "around" your residual target, which i' m not so sure it's acceptable. Achieving the desired residuals is a necessary condition but in some complex cases like conjugate heat transfer the the imbalances must also be maintained low (around 0,1 % is considered very good) for the solution to be considered valid. Along with your diagrams you might another one which shows the imbalances. It could help a more experienced user to help you. Concerning the Courant number, all i know is that it is an indicator on how the flow advances. Large numbers means that fluid particles move rapidly which could lead to divergence. It's directly relevant to the time step size, so reducing it would generally reduce the Courant number.

In order to improve convergence you must do a number of things, if reducing the time step won't suffice. Better meshing, more iterations per time step (solver control, min-max iterations), tighter residual target (1.0e-05-1.0e-05) are among these things, more accurate initial conditions, boundary conditions that might be wrong etc. In complex problems like multiphase conjugate heat transfer flow, you won't believe how much mesh affects the system. I hope you'll figure it out.

edit: i stand corrected by Antanas on the matter of residual targets.

[mohsen.samie]

tnx Antanas and pkladisios. it seems i must reduce timestep and changing the mesh to improving convergenc

[-Maxim-]

Originally Posted by mohsen.samie

in new monitor there is just plot monitor - residual monitor - text monitor and time corrected monitor....what do u mean by imbalances monitor with making new monitor?

new plot monitor and then select imbalances as variables