[Nek5000-users] cyclic boundary conditions and pnpn-2

[nek5000-users at lists.mcs.anl.gov]

Dear Nek users,
I am running both linear and nonlinear simulations of the flow over a  
rotating disk. To decrease the domain size, I use a circular sector of  
the disk which is made possible through the cyclic boundary conditions.

When I use the cyclic boundary conditions for my nonlinear simulations  
together with pnpn-2 they blow up. I am not entirely sure why,  
although, I am currently just using pnpn instead. However, when I use  
these boundary conditions for my linear simulations, I have to use  
pnpn-2 since the perturbation solver is only implemented for this  
specific case, and this seems to work fine. I am also not sure why  
this works fine whereas the nonlinear case blows up (?). Anyway, my  
main question is if there are any plans to develop the pnpn-2 case for  
the cyclic boundary conditions in the nonlinear case? Or maybe what  
I'm wondering is if this would be an easy task to do? We are  
experiencing that it is faster to run with pnpn-2 than with pnpn and  
this would thus be an advantage.

Best regards,
Ellinor

----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.