Lattice Boltzmann methods (LBM) (or thermal Lat- work). It enables complex physics and sophisticated
tice Boltzmann methods (TLBM)) is a class of algorithms. Eciency leads to a qualitatively new
computational uid dynamics (CFD) methods for uid level of understanding, since it allows solving prob-
simulation. Instead of solving the NavierStokes equa- lems that previously could not be approached (or
tions, the discrete Boltzmann equation is solved to simu- only with insucient accuracy).
late the ow of a Newtonian uid with collision models
such as BhatnagarGrossKrook (BGK). By simulating The method originates from a molecular description
streaming and collision processes across a limited num- of a uid and can directly incorporate physical terms
ber of particles, the intrinsic particle interactions evince stemming from a knowledge of the interaction be-
a microcosm of viscous ow behavior applicable across tween molecules. Hence it is an indispensable in-
the greater mass. strument in fundamental research, as it keeps the cy-
cle between the elaboration of a theory and the for-
mulation of a corresponding numerical model short.
1
2 7 SIMULATION OF MIXTURES
proven its potential in computational physics, including 5 Lattices and the DnQm classica-
microuidics: LBM demonstrates promising results in
the area of high Knudsen number ows.
tion
Lattice Boltzmann models can be operated on a number
of dierent lattices, both cubic and triangular, and with or
without rest particles in the discrete distribution function.
A popular way of classifying the dierent methods by
lattice is the DnQm scheme. Here Dn" stands for "n di-
mensions, while Qm" stands for "m speeds. For exam-
ple, D3Q15 is a 3-dimensional Lattice Boltzmann model
4 Development from the LGA on a cubic grid, with rest particles present. Each node
has a crystal shape and can deliver particles to 15 nodes:
method each of the 6 neighboring nodes that share a surface, the
8 neighboring nodes sharing a corner, and itself.[1] (The
D3Q15 model does not contain particles moving to the 12
LBM originated from the lattice gas automata (LGA) neighboring nodes that share an edge; adding those would
method, which can be considered as a simplied cti- create a D3Q27 model.)
tious molecular dynamics model in which space, time,
Real quantities as space and time need to be converted to
and particle velocities are all discrete. For example, in
lattice units prior to simulation. Nondimensional quanti-
the 2-dimensional FHP Model each lattice node is con-
ties, like the Reynolds number, remain the same.
nected to its neighbors by 6 lattice velocities on a trian-
gular lattice; there can be either 0 or 1 particles at a lat-
tice node moving with a given lattice velocity. After a
time interval, each particle will move to the neighboring 6 Lattice units conversion
node in its direction; this process is called the propagation
or streaming step. When more than one particle arrives In most Lattice Boltzmann simulations x is the basic unit
at the same node from dierent directions, they collide for lattice spacing, so if the domain of length L has N
and change their velocities according to a set of collision lattice units along its entire length, the space unit is simply
rules. Streaming steps and collision steps alternate. Suit- dened as x = L/N . Speeds in Lattice Boltzmann
able collision rules should conserve the particle number simulations are typically given in terms of the speed of
(mass), momentum, and energy before and after the col- sound. The discrete time unit can therefore be given as
lision. LGA suer from several innate defects for use in t = x , where the denominator Cs is the physical speed
Cs
hydrodynamic simulations: lack of Galilean invariance of sound.[2]
for fast ows, statistical noise and poor Reynolds number
scaling with lattice size. LGA are, however, well suited For small-scale ows (such as those seen in porous media
to simplify and extend the reach of reaction diusion and mechanics), operating with the true speed of sound can
molecular dynamics models. lead to unacceptably short time steps. It is therefore com-
mon to raise the lattice Mach number to something much
The main motivation for the transition from LGA to LBM larger than the real Mach number, and compensating for
was the desire to remove the statistical noise by replacing this by raising the viscosity as well in order to preserve
the Boolean particle number in a lattice direction with its the Reynolds number.[3]
ensemble average, the so-called density distribution func-
tion. Accompanying this replacement, the discrete colli-
sion rule is also replaced by a continuous function known
as the collision operator. In the LBM development, an 7 Simulation of mixtures
important simplication is to approximate the collision
operator with the Bhatnagar-Gross-Krook (BGK) relax- Simulating multiphase/multicomponent ows has always
ation term. This lattice BGK (LBGK) model makes sim- been a challenge to conventional CFD because of the
ulations more ecient and allows exibility of the trans- moving and deformable interfaces. More fundamentally,
port coecients. On the other hand, it has been shown the interfaces between dierent phases (liquid and vapor)
that the LBM scheme can also be considered as a special or components (e.g., oil and water) originate from the
discretized form of the continuous Boltzmann equation. specic interactions among uid molecules. Therefore,
From Chapman-Enskog theory, one can recover the gov- it is dicult to implement such microscopic interactions
erning continuity and NavierStokes equations from the into the macroscopic NavierStokes equation. However,
LBM algorithm. In addition, th also directly available in LBM, the particulate kinetics provides a relatively easy
from the density distributions and hence there is no ex- and consistent way to incorporate the underlying micro-
tra Poisson equation to be solved as in traditional CFD scopic interactions by modifying the collision operator.
methods. Several LBM multiphase/multicomponent models have
3
been developed. Here phase separations are generated By expanding the particle distribution function into equi-
automatically from the particle dynamics and no special librium and non-equilibrium components and using the
treatment is needed to manipulate the interfaces as in tra- Chapman-Enskog expansion, where K is the Knudsen
ditional CFD methods. Successful applications of multi- number, the Taylor-expanded LBE can be decomposed
phase/multicomponent LBM models can be found in var- into dierent magnitudes of order for the Knudsen num-
ious complex uid systems, including interface instabil- ber in order to obtain the proper continuum equations:
ity, bubble/droplet dynamics, wetting on solid surfaces,
interfacial slip, and droplet electrohydrodynamic defor-
mations. fi = fieq + Kfineq ,
A Lattice Boltzmann model for simulation of gas mix- neq (1) (2) 2
ture combustion capable of accommodating signicant fi = fi + Kfi + O(K ).
density variations at low-Mach number regime has been The equilibrium and non-equilibrium distributions sat-
recently proposed.[4] isfy the following relations to their macroscopic variables
To this respect, it is worth to notice that, since LBM deals (these will be used later, once the particle distributions
with a larger set of elds (as compared to conventional are in the correct form in order to scale from the parti-
CFD), the simulation of reactive gas mixtures presents cle to macroscopic level):
some additional challenges in terms of memory demand
as far as large detailed combustion mechanisms are con- eq
cerned. Those issues may be addressed, though, by re- = fi ,
sorting to systematic model reduction techniques.[5][6][7] i
eq
u = fi ei ,
i
8 Thermal lattice-Boltzmann (k)
0= fi fork = 1, 2,
method i
(k)
0= fi ei .
Currently (2009), a thermal lattice-Boltzmann method i
(TLBM) falls into one of three categories: the multi-
The Chapman-Enskog expansion is then:
speed approach,[8] the passive scalar approach,[9] and the
[10]
thermal energy distribution.
=K +K 2 fort2 (time-scale diusive) t1 (time-scale co
t t1 t2
9 Derivation of NavierStokes
equation from discrete LBE x
=K
x1
.
1 eq
fi (x + ei t , t + t ) = fi (x, t) + (f fi ). fi
(1)
f eq (1) 1 f eq 1 2 fieq f
(2)
f i + i +ei fi + eiei : fieq +ei i + = i .
t1 t2 2 t1 2 t21
For simplicity, write fi (x, t) as fi . The slightly simpli-
ed Taylor series expansion is then as follows, where ":" Then, the second equation can be simplied with some
is the colon product between dyads: algebra and the rst equation into the following:
( ) f eq ( ) [ (1) ]
(2)
fi 1 fi 1 2 fi 1i eq 1 fi (1) fi
+ei fi + eiei : fi + ei + + 1
= t(fi fi ). 2 +
e i 1 f i = .
t 2 t 2 t2 2 t1
4 10 MATHEMATICAL EQUATIONS FOR SIMULATIONS
Applying the relations between the particle distribution where g(x, ei , t) is related to f (x, ei , t) by
functions and the macroscopic properties from above, the
mass and momentum equations are achieved:
(e u)2
g(x, ei , t) = f (x, ei , t),
2
[
]
f j = (f j 0.5)c2s t . (x) = ( ) = 0 1 e( /0 )
6 13 NOTES
Deutsch, Andreas; Sabine Dormann (2004). Cellu- [7] Chiavazzo, Eliodoro; Karlin, Ilya; Gorban, Alexander;
lar Automaton Modeling of Biological Pattern For- Boulouchos, Konstantinos (2009). Combustion simula-
mation. Birkhuser Verlag. ISBN 978-0-8176- tion via Lattice Boltzmann and reduced chemical kinet-
4281-5. ics. Journal of Statistical Mechanics: Theory and Exper-
iment. 2009: P06013. Bibcode:2009JSMTE..06..013C.
Succi, Sauro (2001). The Lattice Boltzmann Equa- doi:10.1088/1742-5468/2009/06/P06013.
tion for Fluid Dynamics and Beyond. Oxford Uni-
versity Press. ISBN 0-19-850398-9. [8] McNamara, G., Garcia, A., and Alder, B., A hydrody-
namically correct thermal lattice boltzmann model, Jour-
Wolf-Gladrow, Dieter (2000). Lattice-Gas Cellular nal of Statistical Physics, vol. 87, no. 5, pp. 1111-1121,
Automata and Lattice Boltzmann Models. Springer 1997.
Verlag. ISBN 978-3-540-66973-9.
[9] Shan, X., Simulation of rayleigh-b'enard convection us-
Sukop, Michael C.; Daniel T. Thorne, Jr. (2007). ing a lattice boltzmann method, Physical Review E, vol.
Lattice Boltzmann Modeling: An Introduction for 55, pp. 2780-2788, The American Physical Society,
Geoscientists and Engineers. Springer. ISBN 978- 1997.
3-540-27981-5.
[10] He, X., Chen, S., and Doolen, G.D., A novel thermal
Jian Guo Zhou (2004). Lattice Boltzmann Methods model for the lattice boltzmann method in incompressible
for Shallow Water Flows. Springer. ISBN 3-540- limit, Journal of Computational Physics, vol. 146, pp.
40746-4. 282-300, 1998.
7
14 External links
OpenLB: LBM community forum for discussions
on research, implementation, open positions and up-
coming conferences, ..
palabos.org: A site with various resources related to
LBM, including a forum.
LBM Method
Lattice Boltzmann summary
Erlangen Lattice Boltzmann mailing list
Entropic Lattice Boltzmann Method (ELBM)
DSFD mailing list: sends information about DSFD
and related conferences, opportunities for doctoral,
postdoctoral, faculty and industry positions for ap-
plicants who have experience with Lattice Boltz-
mann or other related methods.
dsfd.org: Website of the annual DSFD conference
series (1986 -- now) where advances in theory and
application of the Lattice Boltzmann method are
discussed
Website of the annual ICMMES conference on Lat-
tice Boltzmann methods and their applications
8 15 TEXT AND IMAGE SOURCES, CONTRIBUTORS, AND LICENSES
15.2 Images