CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   CFX (https://www.cfd-online.com/Forums/cfx/)
-   -   torque converter simulation (https://www.cfd-online.com/Forums/cfx/112554-torque-converter-simulation.html)

steve ch. January 30, 2013 18:11

torque converter simulation
 
Hello all.
Now, I'm trying to simulate torque converter.
I already set up model but I have no idea how to set up boundary conditions in CFX.
usually, boundary set up in CFX, I have to set up inlet and outlet but in torque converter, it's closed and rotating.
if you have any exemples or hints, please let me know.
thank you in advance.

ghorrocks January 31, 2013 17:06

CFX handles closed domains just fine. You need something to drive the flow, in the torque converter this is impellers whizzing around.

But you should not run a incompressible flow in a closed domain. You will have convergence difficulties. You should add a small port with a pressure control to allow the simulation to "breathe" and control pressure. The torque converter will have a pressure port to do just this - so this is physically realistic.

jsm May 10, 2013 05:11

Might be late reply. But I have question about convergence. How convergence issue could happen in closed domain simulations with in-compressible flows.

In general, oil is used as fluid in torque converter and that is in-compressible in nature. Could you please explain the reason.

Thanks in advance.

JuPa May 10, 2013 05:35

Quote:

Originally Posted by jsm (Post 426487)
Might be late reply. But I have question about convergence. How convergence issue could happen in closed domain simulations with in-compressible flows.

I'm going to take a stab in the dark with this so don't take my answer as the correct solution! My guess is CFX is a pressure-based coupled solver where pressure and velocity is strongly coupled. If you have high velocities and low pressure the solver may not be able to calculate the small pressure differences inside your fluid domain without a "port with a pressure control" as this might allow your domain imbalances to equalise.

ghorrocks May 10, 2013 23:51

There are two main ways of asse3ssing convergence. The residual tolerance is calculated per unit cell as an estimate of how accurately the equations are solved in that unit cell. It does not matter if the simulation has inlets and outlets or not, the units cells do not know that anyway.

The second method is imbalances. This checks the flow in equals the flow out. In closed domains this can be a problem as any imbalance is divided by zero, so even tiny numerical noise becomes a large imbalance. That is why this imbalance tolerance is optional, it is not appropriate for all fows.

jsm May 11, 2013 06:58

Thank you for your replies - Glenn Horrocks & RicochetJ:)


All times are GMT -4. The time now is 16:18.