CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)

 k.vimalakanthan October 16, 2011 14:00

Lift fluctuations in a steady simulation?

1 Attachment(s)
Hi guys,

I'm modelling an airfoil problem at high angle of attack (14 deg) using Ansys CFX (quasi 2D). Currently I'm facing fluctuations in the coefficient of lift value with iterations (please see attached pic). I have tired the simulation with higher physical time scale (even up to 50s), still it does not seems to converge with platoa convergence.

After addressing this issue to some of my friends, they suggest that the simulation is exhibiting a unsteady behaviour in a steady simulation, which i'm not able to comprehend. Is this even possible? :confused:

If this is the case shouldn't my increased physical time scale even out the unsteadiness of the flow?

Any help is greatly appreciated, :)

 truffaldino October 16, 2011 14:22

The steady solution at such aoa is not stable, at this aoa you will have a separation bubble which sheds vortices, just as in case of von-Karman votices around cylinder. Steady solution does not make sence here. You should run transient and then average forces over vortex shedding time period, if you need a mean force.

 k.vimalakanthan October 16, 2011 14:29

Quote:
 Originally Posted by truffaldino (Post 328117) The steady solution at such aoa is not stable, at this aoa you will have a separation bubble which sheds vortices, just as in case of von-Karman votices around cylinder. Steady solution does not make sence here. You should run transient and then average forces over vortex shedding time period, if you need a mean force.
Many thanks truffaldino.

Do you have any idea why changing the local and physical time scales does not effect these 'bouncy' results?
also is it reasonable to average the forces over this periodic steady simulation?

Could you please also suggest any good reference for explaining these 'time marching' CFD techniques?

 truffaldino October 17, 2011 06:12

I think averaging oscillations of steady solver does not make much sence. You should swithch to unsteady. In this situation I usually use implicit scheme with time stepping with courant number somewhere between .1 and 1. Courant number is the ratio of flow travel distance during a timestep to grid spacing C=v*dt/dl, where v-flow velocity, dt-time step, dl-grid step.

 All times are GMT -4. The time now is 05:45.