2nd order accuracy in time using URANS approach

[Alastair West]

I would like an URANS simulation I am currently undertaking to be second order accurate in time. I have altered the necessary variable ISCHTP which automatically triggers second order terms for variables ISNO2T and ISSO2T (source terms of the momentum and energy equations respectively) within the subroutine, varpos.F. However I have encountered this message above the variable ISTO2T (source terms for the turbulent variables): "Termes sources turbulence (k-eps, Rij, v2f ou k-omega) On n'autorise de changer ISTO2T qu'en Rij (sinon avec le couplage k-eps/omega il y a pb)" I forced this ISTO2T variable to be second order (as I am using Menter's k-omega SST) and my results greatly differ from when this variable is left first order.  The result obtained from when this variable is left 1st order is very similar to the result obtained when using an entirely 1st order time scheme with the Courant number well below unity. I understood that the turbulent source terms were coupled only when using the k-epsilon model (ie.IKECOU = 1) and therefore I would not have any problems because I am using the k-omega SST (where by default IKECOU = 0). Could anyone shed some light on the situation? Any suggestions would be greatly appreciated. Regards, Alastair

[David Monfort]

Hello Alastair,
We've been forwarded your e-mail by a colleague, and are discussing your point.
Actually, 2nd order in time has been developped for Large-Eddy Simulation, and adapted for RANS modelling. But 2nd order in time with RANS is not often used... so I need to think a bit before answering.
Sorry for the delay ;)
David