小松弛时间问题怎么解决？

7楼

松弛时间接近0.5时，计算容易发散，这个怎么解决？
是改变格子速度、格子步长？具体是增加，还是减小？

luo@odu.edu

First, abandon the so-called LBGK scheme once for all, and use the MRT-LBE.

Second, if you are doing DNS, then there is a limit on the grid Reynolds number (which is dictated by physics). The point is that stable does not mean correct nor accurate. This point has to be carefully considered.

7楼

谢谢罗老师，但是LBGK自身应该也有解决这种问题的办法吧？

luo@odu.edu

NO -- LBGK cannot control the dissipation of acoustic waves independently, and pressure fluctuations (acoustics) is one major cause of instability.

bitxyq

MRT确实在稳定性上优于LBGK，如果楼主纠结在tau-->0.5计算发散，应果断用MRT，如果你做小Re流动（nu不太小时），LBGK是可以用的。

[ 本帖最后由 bitxyq 于 2013-7-23 14:30 编辑 ]

luo@odu.edu

If you use the LBGK model, YES, it MAY work when you pick tau about 1.0, but you can NEVER reach a converged result (with fixed Ra and Ma).

[ 本帖最后由 luo@odu.edu 于 2013-7-30 06:01 编辑 ]