Multi block interface problem

Hello Dear all
i work on multi block LBM.I impliment Dzhi Yu way explained in his PHD thieses, in chanell flow but there are some problem in my results. and there are apear discontinuty in V,P contor on blocks corner. I used 4 seprate spatial interpolation on each side of block interface first and then tried an overal spatial interpolation on blocks interface. but in both way the discontinuty problem is not solved.If it is possible for any one give me please some guidance clue and help me please to solve this problem.
with the best regard.

I am also confronted with the same problem. For simplicity, i simulate a very simple case, in which there is no any obstacle in field in order to focus my attention on the multi-block interface. But even for this unimaginable simple case, the calculation go wrong. wnen the ratio of multiblock is 1, the result is right. when the ratio is 2, run for tens cycle, the flow is unsymmetrical and at interafce, the vorticity appears.
I wonder how srtmlb and any other guys solve the problem and tell me where is the problem.

I am also try to implement multi-block LBM and I have problem with spatial interpolaction.
Could You give my link to the Dzhi Yu PhD thesis?


I have the same problem (discontinuity at the blocks interfaces )

but Jonas Lätt wrote in his thesis

“In the second part, a modified LB model is introduced under the name of regular-
ized lattice Boltzmann model. In this model,the original LB distribution function
is replaced by aregularized function, inspired by the hydrodynamic terms found
in the Chapman-Enskog expansion. The specificity of the regularized distribution
function is that it can be fully expressed in terms of macroscopic, observable vari-
ables. It can therefore be decomposed into components that are dimensionless in
a system of macroscopic variables, and is rescaled in a controllable way when the
parameters of the space or time discretization vary. Furthermore, as it is shown on
selected CFD applications, the resulting numerical model is more stable and accu-
rate than the original one. The ability to rescale the variables of the regularized
model in a controllable way is exploited in the subsequent chapters. Through a uni-
fied approach, it is explained how to implement grid re?nement, an adaptive time
interval, boundary conditions, initial conditions, and a coupling of the LB model
with another CFD tool.”

I think the reason of discontinuity at the blocks interfaces because of the normal LB model.
So we must use regularized lattice Boltzmann model if we want to rescale the
non equilibrium distribution functions correctly.

actually now I am trying to implement multi block LBM using regularized non-equilibrium
distribution functions.

also if any body can give us some hints I will appreciate it.

best regards



I’m sorry if this statement in my thesis was misleading. I didn’t mean to say that the rescaling works only with the regularized model. As a matter of fact, grid refinement works very well with BGK too, but it’s worth considering the use of the regularized model to improve somewhat the accuracy of the method across the coarse/fine interface.

No matter which model you use, BGK or the regularized model, the crucial point to remember is to rescale equilibrium and off-equilibrium separately, according to their scaling laws.

thanks jonas for your advice