Diverging wall scale due to large domain size
 

I am modelling a blade profile at a Reynolds number of 700000 in air using the SST turbulence model. My solutions are largely domain size dependent and as such i would like to increase the size of my domain to get a more accurate solution. The problem that i am having is that as i make my domain size larger the solver has a tougher and tougher time solving the wall scale. I am at a stage now where if i make my domain any larger the wall scale function goes to infinity and the solver stops. Any suggestions as to how i can prevent thsi problem??

Thank you in advance Kevin

Re: Diverging wall scale due to large domain size
 

Hi Kevin,

The wall scale calculation solves a simple diffusion equation. If it blows up, it is almost invariably due to some bad mesh in your domain. I would check that as you increase your domain size you are not overly distorting the mesh.

Another possiblity, which is easy to check, is if this is due to roundoff error. Try running your case in double precision.

Regards, Robin

Re: Diverging wall scale due to large domain size
 

Thanks for the help Robin. I was wondering how I would go about checking for distortion within my mesh. I have left the meshing technique the same as for previous runs and have only altered the size of the domain. Actually, now that i think of it, i have increased the maximum face spacing in order to maintain a reasonable number of nodal points. Could this be the cause of distortion? My smallest elements at the surface of the blade are approx. .01 mm while at the boundary of the domain they reach a size of approx. 500 mm and I was wondering if this could be causing the complications. Finally, i am not entirely sure how to set the solver to use double precision and i was hopng that you could point me in the right direction.

Thank you again, your support and knowledge is always appreciated.

Kevin

Re: Diverging wall scale due to large domain size
 

Hi,

The large range of mesh element sizes is potentially causing a problem with the single precision solver I strongly recommend you try double precision.

You can run the double precision solver by going to the advanced options on the solver manager when you start a run up, or if using the command line you can use the switch -double

Regards, Glenn Horrocks