CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   NACA0012 Convergence Problems (http://www.cfd-online.com/Forums/cfx/21844-naca0012-convergence-problems.html)

StudentAndrew November 18, 2005 06:18

NACA0012 Convergence Problems
 
I am currently running simulations for a NACA0012 aerofoil at various angles of attack and am having problems with convergence, currently at AoA=10deg, in 15m/s flow. I am running the simulation as a steady state problem for 70 or so iterations with a local timestep factor of 5. I am then using these results to perform a transient simulation. I have tried a few different combinations of time steps, total times and iterations per timestep (for eg 0.1s 10 iterations per step for 5s up to 2s with 5 iterations per step for 80s) but the problem still won't converge! My convergence criteria is only 1e-04 RMS which I wouldnt think would be too difficult to achieve, especially at a relatively low AoA. I am using the SST turbulence model with turbulence Eddy dissipation. Does anyone have any ideas as to where I may be going wrong?

Any suggestions will be much appreciated!!!

Thanks,

Andrew

StudentAndrew November 18, 2005 06:22

Re: NACA0012 Convergence Problems
 
What is happening, by the way, is that the simulation is tha the convergence graph is oscillating (which I'm I guess it should be doing for a transient run) but not diverging at the same time to hit 1e-04, so the simulation is stopping when the total time is reached, rather than the criteria is met.

Thanks!

Andrew

longbow November 18, 2005 10:43

Re: NACA0012 Convergence Problems
 
To figure out the cause of convergence problem, you probably want to identify where is the maximum residual located. This will give you an idea whether the problem is caused by bad mesh quality or transient phenomena, like separation or vortex shedding.

James Date November 18, 2005 11:58

Re: NACA0012 Convergence Problems
 
Its worth seeing if you can get a nice converged solution using the k-e turbulence model then once you have that cracked move onto the SST. Even use the k-e converged solution as the intial guess. I found that using the correct y+ value was the key to getting good convergence. Its always with starting with a coarse grid and getting good convergence and building up to a high resolution mesh. I did a lot of work solving flows around the NACA0012 a few years back. If you want a copy of my results, give me a shout and i'll pdf them to you.

StudentAndrew November 20, 2005 13:10

Re: NACA0012 Convergence Problems
 
Hi James,

Thanks for the advice. Your NACA0012 data would probably be really useful if you could send it to me!

Thanks again,

Andrew

DAK565656 November 21, 2005 06:41

Re: NACA0012 Convergence Problems
 
James, were there any problems with drag? Have you obtained good results?

James Date November 21, 2005 07:49

Re: NACA0012 Convergence Problems
 
Andrew, send me your email address and i'll send you a copy of my results. The reference for this work is: "Performance prediction of high lift rudders operating under steady and periodic flow conditions", PhD Thesis, University of Southampton, 2001, James Date, British Library location: DSC DXN049180

DAK - I found a slight over prediction in drag at low AoA using the both the standard an RNG k-e models. This was probably due to the laminar/turbulent transition not being accounted for.

James


All times are GMT -4. The time now is 00:34.