Hi, I'm wondering if anybody out there can help me. I am looking for information regarding the "blended Second Order Upwind/Central Difference" discretisation scheme. I am trying to compare my work on mass transport to previously published work which uses this scheme but Fluent have discontinued this upwinding technique due to "numerical instablilities".

When I compare my computational results to the published data, both the 1st order and Power law schemes provide good agreement. However, using the higher, more accurate schemes (2nd order, QUICK) there is large differences between the computaional and published data.

I am wondering if the "blended Second Order Upwind/Central Difference" tends to the 1st order scheme in highly convective flows in a similar fashion to the Power Law scheme? I would also like to know what "numerical instabilities" occur in the "blended Second Order Upwind/Central Difference" and how they come about.

Any advice regarding this would be greatly appreciated as I've tried looking for information on "lended Second Order Upwind/Central Difference" but no CFD book I've looked at disccuses it.

Thanking you in advance,

Paul

Hi, if you are referring to first order upwinding with central differencing using "donor cell factor" (hybrid discretization scheme or higher order upwinding) then you should be able to control instabilities that arise due to upwinding by choosing a proper donor cell factor. central difference is generally valid for low Re flows but has good accuracy. for highly convective flows, either QUICK or hybrid discretization scheme should generally work. it seems OK superficially to assume that upwinding properties (it is also first order accurate) should prevail at high Re but may result in instabilities in herent in its definition.

Regards, Swarup