CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   Inlet Motion with 2-way FSI (

browns6 January 9, 2013 15:59

Inlet Motion with 2-way FSI

I am running a 2 way FSI simulation of a pipe. The inlet is specified to move in the x, y and z directions with zero radial rotation. The inlet flow is time varying, (uniform profile for one simulation and uvw velocities for the second simulation). I am having problems with convergence of the FSI model. Both the fluid model and the shell converge separately, but not when linked together. Length of the pipe runs axially in the z direction with xy being the inlet plane.

There is a warning: "In Analysis 'Flow Analysis 1' - Domain 'Default Domain': Mesh motion is specified on the boundary 'Inlet'. This is valid for motion that is tangential to the boundary but solution errors will occur if motion is normal to the boundary." that occurs in the FSI set up.

Does this warning mean that if the inlet moves in xyz directions and the flow is normal to the inlet, there is no way a solution can be obtained for the model?

I have run the model previously with just z inlet motion and it converges with no errors.


ghorrocks January 12, 2013 16:25

You can still do the simulation, but the error message says the normal motion component will not be taken into account. It will converge fine and report no errors, but may not be correct... but may be close enough - that depends on what you are simulating.

KeganLeckness May 27, 2016 11:36

Hi all,

I have a similar set-up. I'm doing a 2-way FSI simulation of pipe flow. I need the inlet to translate in the normal direction. I cannot prescribe this motion via UDF, however, because the motion of the pipe (and subsequent motion of the inlet) is unknown ahead of time. How may I define the the inlet to move freely with the structure?

Currently, the outer nodes of the inlet are in motion with the structure, but the rest of the inlet remains stationary - shearing the fluid elements and stopping the solution.

Any advice is appreciated. Thanks!

ghorrocks May 28, 2016 05:55

If you have fully developed pipe flow going in then you can probably prescribe the velocity profile. The next bit is speculation: You might be able to work out the mesh motion component and add that to the inlet velocity to get the relative inlet velocity. I don't know if this helps but it might be worth a try to see if reduces the moving inlet problem.

But before you do anything I would do a simulation with a moving inlet by itself (maybe simple pipe flow only with no FSI) so you can compare against analytical results. Then you can see if the problem is significant enough to do something about.

All times are GMT -4. The time now is 21:46.