Interesting result from oscillating cylinder
hi,
i've been trying to simulate an oscillating cylinder flow using ALE with fractional step. i can't get the correct results although i can't find any error from the debugging. then i refine the grids in the normal direction from 40 to 80 and volia! i finally get the correct results! the interesting thing is that my past unsteady simulation with static/rotating cylinder using the old grid size give very good result, hence i don't feel the need to increase the grid no. It seems that using less grid, the "lock on" is not possible. Is my reasoning correct? I didn't know that an increase in grid can give such different answer... even when previous simulation already give good results... |
Re: Interesting result from oscillating cylinder
btw, does anyone knows how to speed up the process to achieve the "lock in" stage? i've to run around 8000 to 9000 step (equivalent to non-dimension time of 100+) to reach it.
Tks |
Re: Interesting result from oscillating cylinder
That's interesting and strange. How large are the cylinder oscillations (in terms of diameters) and how different is the oscillation frequency from the steady vortex shedding frequency?
One thing to pay attention to: your cylinder is moving and if the grid is rigidly attached to the cylinder, it will sweep across the wake twice for each cycle. Each sweep numerically diffuses the wake, and that can have an impact on the solution, if the grid is too coarse. How do you non-dimensionalize time (you should use time/(oscillation period))? |
Re: Interesting result from oscillating cylinder
the oscillation ampltitude is 0.2 of the diameter. when the grid is coarse, the shedding freq remains the natural one and the cl is a sine wave with varying amplitude.
however, when the grid is refined, the "lock on" occurs and the shedding freq becomes equal to he oscillating freq. the cl also changes to a sine wave with fixed amplitude. |
Re: Interesting result from oscillating cylinder
how close is the structural natural frequency to the static vortex shedding frequency? it is quite possible that the lock-on phenomenon is sensitive to numerical dissipation, hence the need for a fine grid. This sensitivity should decrease as you move the structural frequency closer toward the vortex shedding frequency. I would survey the literature for evidence on that. For example:
http://arjournals.annualreviews.org/....050802.122128 If you don't find anything conclusive in the literature, you may consider performing a more detailed analysis, which might then be worth publishing. |
All times are GMT -4. The time now is 16:59. |