I would like to use UDS to calculate the concentration field. But the results are negative (small) in some parts of the domain. Would it affect a lot on the solution? and if that's true, how can I solve a positive scalar by uds? Can I just set the negative values to zero using udf?

Thanks a lot for your advising.

Jun Li

There are couple of things to do.

Generally I check for small negative values and set them to zero inside a DEFINE_ADJUST routine. Scalars should never be negative (see Patankar).

The other option, which you should also consider is to add or modify your source term - Patankar also describes this - so that it is linearised with respect to the scalar being solved and forces the solution to zero as zero is approached.

In practice even with some sort of source term linearisation I find that I still need to reset small negatives to zero. gnerally we are talking about numbers of order 1.0e-4 to 1.0e-6 in magnitude so they don't effect the solution if reset to zero. However it you leave them they may grow exponentially and interact with your other equations in an adverse way, so its best to reset them to zero.

Hope it helps

Greg

ps Patankar - Numerical heat Transfer 1980 - the first book to read if you're new to CFD and finite volume methods.

I appreciate it, Greg. I am doing it now and will let people here know what will happen when setting the negatives to zero.

Jun

No much change happens in my case if resetting the negatives to zero. Jun