Transcript qmp

1
Mesoscale modelling with
the Lattice Boltzmann model
Jonathan Chin
<[email protected]>
Lattice Boltzmann Method
• Fluids represented by fictional “particles” with discrete
velocities, occupying sites on a discrete lattice.
• Particles described by real-valued distribution
function
• Density of single component  given by
  (x)  m  f i (x)
• Momentum of single component
given by
i



 u m

f
 i ci
i
Time Evolution
• Advection step: particles move to adjacent sites.
• Collision step: distribution function relaxes to
equilibrium value via BGK operator.


fi '  fi 
1

f

i
 f i (eq) 
• The equilibrium distribution is a function only of
density n and velocity uat a given site.
fi
(eq)
 ci ci u u
 nTi 1  2 
2
cs
2 cs

 ci ci

 2    
 cs

• Weighting factors Ti chosen to ensure isotropic
hydrodynamics.
• Bounce-back boundaries give non-slip walls.
• Velocity is perturbed to take account of collisions with
other components, and forcing term.










v     u  /       F 
 
    
• Force term includes gravity and Shan-Chen
immiscibility force.


ˆ
F (x )  g gravn z  n (x ) n (x  ci )ci



• Immiscibility force proportional to single-component
density gradient, repelling differing components.
Spinodal decomposition
• Two fluids may mix above some temperature
TC
• A mixture quenched below this temperature will
separate into its component fluids: this process is
called Spinodal Decomposition.
• Simulation performed of demixing process using
periodic boundary conditions.
Phase separation images
Timestep 0
500
1000
4000
8000
32000
2000
50000
Growth Regimes
• Very early time: exponential growth in structure factor
as interfaces form according to a Cahn-Hilliard model.
• Early-time hydrodynamic regime dominated by
viscosity.
• Late-time inertial hydrodynamic regime: domains grow
as two-thirds power of time.
• Very late time turbulent mixing regime postulated but
not seen.
Early-time Cahn-Hilliard Growth
• Circularly-averaged
structure factor S(k) retains
shape but grows
exponentially in magnitude
as domains form.
• Although the model does not
define any free energies, it
produces results in
agreement with free-energy
models of phase separation.
Domain Growth Laws
Porod Law structure
Power law growth
Breakdown of scaling
• Many theories assume that a phase-separating system
contains a single length scale evolving in time.
• Simulation shows regime with multiple length scales due to
competition between different growth mechanisms.
Surface tension measurement
• “Laplace’s Law” states that the
pressure drop across the
interface of a bubble is
inversely proportional to its
radius.
p 

R
• Bubbles simulated with
different coupling constants to
measure surface tension.