why LBM?


The title says it all,
I start working with LBM but I can’t find why using this method and not another. My chief say that it’s easily for programmation and dealing with complex boundary conditions.
But in terms of computation time and accuracy (and so cost) I have no idea!
Can I get information about it?
thank you

There are several reasons why a user may want to choose LBM over other CFD methods, and also several against it (note, for example, that the LBM is usually inefficient for time independent flows). There are a number of articles which make their arguments either for or against but it can sometimes be difficult to work through the propaganda. Perhaps one of the more convincing reasons for is most traditional methods have to discretise a complicated non-linear convection term in the Navier-Stokes equations (u.grad u). The LBE, on the other-hand, comes from the discrete Boltzmann equation (ie the Boltzmann equation with a finite set of particle velocities) - this is a linear constant coefficient hyperbolic PDE. In other words, it replaces non linear convection with linear, constant coefficient advection. All the nonlineartites are bundled into an algebraic source term, which is local in space and time. Succi has quite a nice slogan: “Nonlinearity is local, non-locality is linear.” This makes the method amenable to high performance computing on parallel architectures, including GPUs.