[taw]

Dear sirs,

I am a bit intrigued about the use of under relaxation factor in the solution of the NS equation obtained by pressure correction method.

I am familiar about the need underrealxing, after every inner iteration, for instance p=p+\alpha p'. One convincing reason is that , due to the linearization of the inherently non-linear NS equation, we have to update the new velocity / pressure after underlaxing the new solution. The other reason could be related to the formation of the pressure correction (PC) equation. Since PC ignores some terms from the momentum equation while it is constructed. So due to this inconsistency, for some problem the pressure can be big, so we need to come up with a reasonable underrelaxation factor that match the specific problem.

Having said the above and assuming I am rigfht so far, my question is :

Do we use underelaxation method for DNS like simulation, which often use fractional step method for coupling the velocity with the pressure. Since I often come across the issue on underelaxation discussed in relation with steady state or turbulence modeling only.

Will DNS simulation still be time accurate if underelaxation is used.


Regards,

TAW