[ripperjack]

Hi guys,

I would like to calculate the div(phi,U) term using both 1st order scheme (i.e., Gauss upwind) and 2nd order scheme (i.e., Gauss linear) in my simulation. I know that I can define separate fields such as phi_2nd and U_2nd, and phi_1st and U_1st, and then I can set the numerical schemes for these fields separately. But if I do it this way, the memory usage will be doubled. Is there a convenient way I can do it without defining new fields? Thank you guys in advance!

Best regards,
Jack

[ripperjack]

Originally Posted by ripperjack

Hi guys,

I would like to calculate the div(phi,U) term using both 1st order scheme (i.e., Gauss upwind) and 2nd order scheme (i.e., Gauss linear) in my simulation. I know that I can define separate fields such as phi_2nd and U_2nd, and phi_1st and U_1st, and then I can set the numerical schemes for these fields separately. But if I do it this way, the memory usage will be doubled. Is there a convenient way I can do it without defining new fields? Thank you guys in advance!

Best regards,
Jack

Hi guys,

I am very close to getting this fixed. I can read and modify the fvScheme IOdictionary at run time. However, even I change the divscheme of div(phi,U) from Gauss linear to Gauss upwind, the results do not change... I think I need to call a function to let fvc::div know that I want to use the updated div scheme. Any ideas? Thanks!

Regards,
Jack