[saeedi]

Hi,

I need some help on a problem.

I am solving the scalar transport equation which is fully decoupled from the momentum (passive scalar).

For a specific type of BC (very high gradient around a small area), many schemes are not stable. Thus, I have to use a stable method. The most stable scheme is of course simple up-winding but it is first-order and very dissipative.

I know that I can increase the accuracy and decrease the false diffusion through introducing some extra terms (from the old time step) to the source terms know as deferred correction. I just don't know how exactly I should add them.

I read some in chapter 5 of the book of Versteeg about TVD schemes, but wanted to know if there is any other type of deferred correction which only uses one neighbour points in each direction.

Thanks in advance for your help.