QU KUN
|
February 23, 2004 03:46 |
Re: How much do I numerical dissipation?
The scheme is not of FDM/FVM/FEM. It is a kind of lattice boltzmann scheme in which viscosity can be controlled by a parameter Tao. At every node, the viscosity nu is a function of pressure P and collision time Tao. If the viscosity is known, I can get Tao according to P. Now I don't know how to estimate the best viscosity nu. If nu is too small, Tao will be too small and the solver will diverge. If it is too large, Tao will be too large and the result will be absurd.
|