CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   Main CFD Forum (https://www.cfd-online.com/Forums/main/)
-   -   high frequency errors (https://www.cfd-online.com/Forums/main/15235-high-frequency-errors.html)

sudhakar May 28, 2008 00:51

high frequency errors
 
hi, I have simulated the flow field over oscillating circular cylinder in free stream flow. The problem is I am getting high frequency oscillations in my cd plot. Can any one tell me how to eliminate those errors? Thank you very much..

Harish May 28, 2008 15:40

Re: high frequency errors
 
I am not sure of the numerical discretization that you use but since the oscillations are high frequency parasite waves I assume its some form of central differencing. There are two ways to remove the oscillations. The first way is to add dissipation/damping terms(which you can find in any CFD book, jameson dissipation , papers of Tam) to the equations and the second way is to use numerical filtering of the solution( Look for paper by Lele on compact scheme, papers of visbal and gaitonde).

sudhakar May 29, 2008 01:30

Re: high frequency errors
 
hi harish, Thanks for your kind reply. I am discretizing convective terms in conservative form using central diff scheme; because if I use upwind schemes, I am getting higher drag value. I will try use numerical filtering. thanks once again..

skg May 29, 2008 04:29

Re: high frequency errors
 
Hi sudhakar, You hav'nt written the in which Reynolds number,grid quality(fine or coarse?),frequency etc. you are using. In case of oscillating cylinder in free stream flow the two main parameters influence the flow are i) forcing frequncy of oscillation ii)amplitude of oscillation. The extra frequencies may not come from discretization error as you are using central differencing(CD)+Upwinding(Order?), which helps to dampen the oscillation property of CD. On the other hand this hybrid scheme accuracy of the discretization may bring down to first order. Presence of extra frequency component in lift spectrum carries another meaning of flow instability.

sudhakar May 29, 2008 08:49

Re: high frequency errors
 
hi skg, Thanks for your reply. The parameters describing the flow conditions are: oscillation frequency=0.154, amplitude=0.2*diameter and Re=185. Also I am not using hybrid scheme. I use CDS only. In literature, a smooth cd curve is presented, but my Cd vs time curve oscillates with high freq, eventhough the avg cd is predicted with good accuracy. Any suggestions?

skg May 30, 2008 01:58

Re: high frequency errors
 
Hi, sudhakar I can't understand ur amplitude parameter. I have assumed that the cylinder is oscillating in sinusoidally. In that case u have to give ur amplitude parameter in terms of angular displacement i.e. in terms of PI(PI=3.14159...). I think u have run ur programme as flow past stationary circular cylinder. Compare ur Cl,Cd values with the standard bench mark results. Also compare ur results varying the grid parameters. CDS may produce oscillation if the grid is too coarse. At the same time CDS in fine uniform grid exibits second order accurate results. How u are calculating the frequencies of Cd curve and how long in non-dimensional time parameter ur programme have been excuted? If every thing is OK, short time execution of programme may furnish some undesired high frquencies.

sudhakar May 30, 2008 03:34

Re: high frequency errors
 
hi skg, My cylinder is oscillating in vertical direction sinusoidally (it does not rotate).I run for 20 oscillation cycles. I think I have to reduce my time step. Thanks for your kind reply..


All times are GMT -4. The time now is 14:38.