openlb looks quite interesting and I wonder, if it is suitable for flows in a diffuser with a pressure driven separation, 3D wall jets and problems with natural convection (e.g. a heated wall)? It would be great! If so, does anyone could give a hint, how to create a mesh of r a 3D diffuser (e.g. http://220.127.116.11/ercoftac-sig15/case132.html ) for openlb?
Does there exists any special wall treatment in openlb or does one have to refine the mesh at the walls just like in LES run using Navier-Stokes?
I do have the feeling that you are looking for a full-featured, general purpose CFD framework, which OpenLB is not. As an answer to your question, it is indeed possible to do all this with OpenLB. However, the library has been developed mainly for research, and it is unlikely that you can obtain significant results without a certain understanding of the lattice Boltzmann method.
This being said, I would suggest that you have a look at the example programs to understand how to set up a simulation. In particular, there is a program which illustrates how to set up a problem with natural convection.
no, actually I am not looking for a full featured code, but would rather like to know, if LBM/openlb is suitable for these kind of flows. I could not find too much literature about flow cases with a large influence of the walls. I would give it a try with openlb for the mentioned generic cases, but if is not recommend at all, I cancel it!?
There is a module for doing thermal flows in OpenLB. At the current stage, the main limitations are:
It is based on the Boussinesq approximation (temperature acts on the velocity by means of a linear buoyancy term).
There is no LES model, which means that you cannot reach highly turbulent regimes. On an “average” parallel machine and in 3D simulations, you should not expect to be able to exceed a Reynolds number of around 10’000.
There is, so far, no implementation of inclined walls. This means that all walls must be parallel to the main axes (x-, y-, or z-axis).
If you can accept these limitations, you will probably find that OpenLB is easy to use and does a good job to model the physics of your problem. You may want to have a look at the following simulations, which were all done with OpenLB. In particular, the simulations of the Rayleigh-Beynard convection serve as a quantitative validation of OpenLB for thermal flows, as they show that one recovers the critical Rayleigh number with high accuracy.