![Page 1: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/1.jpg)
![Page 2: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/2.jpg)
Microscopic picture:
-coordinates and velocities of all
particles, cars, cells, rabbits…
Systems of ODE - ordinary differential equations
Direct simulation methods Monte Carlo methods
Macroscopic picture: average velocity field or density field depending on position
Navier - Stokes equation diffusion equation Shallow water equation
Mesoscopic picture: - “number” of particles
with velocities and positions Boltzmann equation Lattice – Boltzmann equation neutron transport
![Page 3: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/3.jpg)
![Page 4: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/4.jpg)
Distribution function
Information is required not only about the spatial distribution of molecules but about their velocities as well.
![Page 5: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/5.jpg)
Distribution function and macroscopic parameters
- concentration
- density
- velocity
- energy
- peculiar velocity
![Page 6: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/6.jpg)
Distribution function and fluxes
![Page 7: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/7.jpg)
Distribution function and fluxes
- pressure tensor
- normal stresses
- shear stresses – force per unit area in direction exerted on a plane surface with normal in direction
-hydrostatic pressure
- heat flow vector
![Page 8: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/8.jpg)
Transport in absence of collisions
![Page 9: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/9.jpg)
The Boltzmann kinetic equation
The Knudsen number Kn shows the relation between the collision frequency and the typical macroscopic time scale. In the limit of Kn 0 one observes the macroscopic behavior.
(1844-1906)
![Page 10: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/10.jpg)
The Boltzmann collision operator (terrible formulas!)
![Page 11: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/11.jpg)
Collision invariants
Collision invariants are functions of particles velocities that are preserved during collisions. Typical examples are: mass m , momentum and kinetic energy . These conservation laws are expressed as:
or
![Page 12: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/12.jpg)
Collision invariants: why these?
![Page 13: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/13.jpg)
Maxwell –Boltzmann equilibrium distribution
has two important properties: it does not change under collisions
It minimizes the H- function for fixed macroscopic velocity, temperature and concentration. Functional is called entropy
![Page 14: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/14.jpg)
Macroscopic equations
To see how macro-parameters depend on time and position we use conservation laws corresponding to our system. We multiply the kinetic equation
by collision invariants and integrate it over velocities. Such an operation is called scalar product of functions in analysis. The integral over collision operator disappears and the rest gives us macroscopic equations. One says that the collision operator is orthogonal to collision invariants.
![Page 15: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/15.jpg)
Macroscopic equations
![Page 16: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/16.jpg)
Macroscopic equations
![Page 17: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/17.jpg)
How to find macroscopic fluxes corresponding to kinetic models? (and vise versa…)
![Page 18: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/18.jpg)
BGK model
BGK model keeps most of formal properties of the Boltzmann equation: the same collision invariants, the same equilibrium, H – theorem, but has much simpler structure. Knudsen number Kn shows the relation between collision frequency and the typical macroscopic time scale. In the limit of Kn 0 one observes the macroscopic behavior.
![Page 19: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/19.jpg)
Chapman - Enskog method and method of diffusive scaling
(…hard calculations!)
David Enskog Sydney Chapman
(1916, 1918) (1917, 1922)
![Page 20: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/20.jpg)
Transport coefficients:
Viscosity, heat conduction and diffusion coefficients of real gases are rather complicated functionals of particular collision models (intermolecular potentials). The theory becomes even more complicated for mixtures of gases.
The Chapman-Enskog theory shows that in the limit of Kn --> 0 density, velocity and temperature: satisfy up to the second order in Kn number, Kn0 classical macroscopic equations - Navier Stokes equations for compressible viscous gas. It is remarkable that this conclusion is valid both for the full Boltzmann equation and for the simplified BGK equation.
![Page 21: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/21.jpg)
Lattice Boltzmann equation
![Page 22: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/22.jpg)
Distribution function for discrete velocity space:
![Page 23: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/23.jpg)
Similarly the stress tensor is calculated:
![Page 24: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/24.jpg)
![Page 25: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/25.jpg)
![Page 26: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/26.jpg)
![Page 27: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/27.jpg)
![Page 28: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/28.jpg)
Lattice Boltzmann for advection – diffusion
![Page 29: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/29.jpg)
![Page 30: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/30.jpg)
Lattice- Boltzmann algorithm: -transport
![Page 31: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/31.jpg)
Lattice- Boltzmann algorithm: -macroparameters
![Page 32: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/32.jpg)
Lattice- Boltzmann algorithm: -relaxation
![Page 33: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/33.jpg)
Boundary conditions
![Page 34: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/34.jpg)
Boundary conditions for flow
1) Bounce back boundary conditions imply zero (non-slip) velocity for flow and
zero total flux for diffusion. 2) Fixed pressure (for flow) or density (for
diffusion) 3) Slip boundary conditions for flow: friction
depending on tangential velocity 4) Zero normal flux of solute particles (no
absorption) 5) Robin boundary conditions: normal flux is
a function of the concentration on the wall (interaction with the structure)
6) Mirror and periodic boundary conditions.
![Page 35: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/35.jpg)
Neumann (zero normal flux) boundary conditions for diffusion
We wish to have a boundary condition of the type i.e. zero flux across the boundary. We need to assign values for the distribution function for the incoming velocities that intersect the boundary. We assign an equilibrium distribution at the solid nodes:
![Page 36: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/36.jpg)
The unknown density and flux are computed using a mirror point using a mirror rule for the flux Values are interpolated to the mirror point from neighboring grid nodes. Normals nj and distances qj are averages from intersection points.
Neumann (zero normal flux) boundary conditions for diffusion
![Page 37: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/37.jpg)
Example: flow through a gel structure
![Page 38: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/38.jpg)
![Page 39: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/39.jpg)
Parallel computations by domain decomposition
![Page 40: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/40.jpg)
Diffusion LBE. Diffusive scaling limit. (Hard calculations!)
To make notations shorter we skip indexes after the discrete velocities of particles and indexes after for distribution functions and write instead. Diffusive scaling means that we choose time scale:
The differential equation corresponding to the diffusive scaling is
To model diffusion we introduce the equilibrium function as:
We introduce the asymptotic expansion to see the connection between the Lattice Boltzmann equation and the diffusion equation for small = and will compute corresponding diffusion coefficient.
![Page 41: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/41.jpg)
Diffusion LBE: continuum case.
We put the expansion into the equation
and collect terms of the same order, that gives equations for :
![Page 42: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/42.jpg)
Diffusion LBE: continuum case.
![Page 43: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/43.jpg)
Diffusion LBE: continuum case.
![Page 44: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/44.jpg)
First order terms in the expansion
![Page 45: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/45.jpg)
Diffusion equation as a compatibility condition of second order
,
![Page 46: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/46.jpg)
Diffusion LBE: discrete case
![Page 47: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/47.jpg)
Diffusive limit. Discrete case (similar)
- diffusion equation
- diffusion coefficient
![Page 48: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/48.jpg)
Diffusion through EC/HPC(30%) film
Example of diffusion
![Page 49: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/49.jpg)
Flow through a sample of EC/HPC(30%) film
Example of flow
![Page 50: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/50.jpg)
Diffusion and flow streamlines
As a result of computations we get a velocity field for flow, and a flux field
Streamlines for diffusion Streamlines for flow
for diffusion and for advection – diffusion. Velocity here is zero in the pure diffusion case. Streamlines are computed by integration of ODE along these vector fields and show the average paths of fluid particles for the flow case and solvent particles in the diffusion case.
![Page 51: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/51.jpg)
Examples of multyphase flows.
Three phase flow Two phase flow and diffusion of the third phase
![Page 52: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/52.jpg)
Multy-phase flows. Shan Chen model
- is common velocity of phases
is the sum of forces acting between different phases
is responsible for phase separation
is responsible for adhesion forces
![Page 53: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/53.jpg)
Interphase forces in the Shan Chen model
is the sum of forces acting between different phases
For simple repulsion between different species the interactive force in a binary mixture can be chosen as:
where and mark different phases.
Similarly the adhesion between fluid particles and the solid boundary can be chosen :
Where is 1 if the node i is inside the solid and zero otherwise.
To model a mixture of wetting and non-wetting fluid can be chosen
positive for wetting fluid and - ( negative with the same absolute value)
for non-wetting fluid.
![Page 54: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/54.jpg)
Macroscopic equations corresponding to the Shan Chen model of multyphase flows
![Page 55: Lattice Boltzmann simulation - Chalmers · Direct simulation methods Monte Carlo methods Macroscopic picture: average velocity field or density field depending on position Navier](https://reader033.vdocuments.site/reader033/viewer/2022042908/5f39491f97ea375522489f4d/html5/thumbnails/55.jpg)
The parameter defines missible or immissible character of multiphase flows in the Shan Chen model. The thickness of the interface increases with decreasing and fluids mix together below certain threshold. The difference of pressure inside and outside of a bubble satifies the Young – Laplace law : it is proportional to the curvature of the bubble 1/R.