Hello, I am a LBM new learner. I am computing 2D flow past a cylinder,
I got the code from this forum and I noticed that that code used
bounceback rule to wall nodes. Now I intend to implement more
accurate boundary treatment like inamuro or Zou-he ,regularized,
etc. I have a question that if I use these boundary treatment, if
the collision step is needed before streaming?
I am just confused by the standard procedures when coding my lbm
program, it seems that many articles focus more on theory but less
on how to implement, which is difficult for me to get started.
I had the very same problem as you when starting out (only been using this for a month or two). There is a LOT of help on this forum, i found it useful to simply go through some of the existing posts and read what people had been discussing.
To answer your question, you do need the collisions step, it is part of the lb method. The different types of boundary condition you have specified are for domain boundaries (not internal no-slip boundaries), there existence are attempts at solving various problems associated with the domain boundaries. One such problem is the accumulation of mass with prescription of a boundary velocity. Someone linked this paper which i have found to be quite interesting/useful: http://www.lbmethod.org/literature:latt_08
Thanks, brucedjones, do you mean these boundareis can not be used to the internal no-slip wall and if I want to treat no-slip wall,
I have to use bounceback rules but no other ways? As I know, inamuro and Zou-he BC could also be used to no-slip wall, so
I am getting more confused.
You are correct. The issue I think with Inamuro and Zou-He is that they can only be implemented on straight boundaries, I may be wrong however, its always best to confirm these things for yourself.
edit: after thinking about it you may be able to use those BC’s within the domain so long as they are straight… not sure what the implications of doing as are, I haven’t tried it.