concentration divergence for large Scmidt numbers

Philip

I'm trying to model a flow with a passive scalar, but I have trouble when the Schmidt number is large.

For Schmidt numbers above 10, or diffusion coefficients below 10^-4, the concentration of the passive scalar diverges (I'm using unstructured mesh with second order upwind or third order muscle).

Is there a way to perform these calculations without divergence?

Is it possible to implement a flux limiter to the third order muscle scheme through a UDF file?

Regards Philip