Homec4science
Diffusion lammps (lammm-devel) Browse

lammps/doc/fix_lb_fluid.html
1cada0b49692lammm-devel

Last Change
Raw File

fix_lb_fluid.html
View Options

<HTML>
<CENTER><A HREF = "http://lammps.sandia.gov">LAMMPS WWW Site</A> - <A HREF = "Manual.html">LAMMPS Documentation</A> - <A HREF = "Section_commands.html#comm">LAMMPS Commands</A>
</CENTER>
<HR>
<H3>fix lb/fluid command
</H3>
<P><B>Syntax:</B>
</P>
<PRE>fix ID group-ID lb/fluid nevery LBtype viscosity density keyword values ...
</PRE>
<UL><LI>ID, group-ID are documented in <A HREF = "fix.html">fix</A> command
<LI>lb/fluid = style name of this fix command
<LI>nevery = update the lattice-Boltzmann fluid every this many timesteps
<LI>LBtype = 1 to use the standard finite difference LB integrator,
2 to use the LB integrator of <A HREF = "#Ollila">Ollila et al.</A> &#91;1&#93;
<LI>viscosity = the fluid viscosity (units of mass/(time*length)).
<LI>density = the fluid density.
<LI>zero or more keyword/value pairs may be appended
<PRE>keyword = <I>setArea</I> or <I>setGamma</I> or <I>scaleGamma</I> or <I>dx</I> or <I>dm</I> or <I>a0</I> or <I>noise</I> or <I>calcforce</I> or <I>trilinear</I> or <I>D3Q19</I> or <I>read_restart</I> or <I>write_restart</I> or <I>zwall_velocity</I> or <I>bodyforce</I> or <I>printfluid</I>
<I>setArea</I> values = type node_area
type = atom type (1-N)
node_area = portion of the surface area of the composite object associated with the particular atom type (used when the force coupling constant is set by default).
<I>setGamma</I> values = gamma
gamma = user set value for the force coupling constant.
<I>scaleGamma</I> values = type gammaFactor
type = atom type (1-N)
gammaFactor = factor to scale the <I>setGamma</I> gamma value by, for the specified atom type.
<I>dx</I> values = dx_LB = the lattice spacing.
<I>dm</I> values = dm_LB = the lattice-Boltzmann mass unit.
<I>a0</I> values = a_0_real = the square of the speed of sound in the fluid.
<I>noise</I> values = Temperature seed
Temperature = fluid temperature.
seed = random number generator seed (positive integer)
<I>calcforce</I> values = N forcegroup-ID
N = output the force and torque every N timesteps
forcegroup-ID = ID of the particle group to calculate the force and torque of
<I>trilinear</I> values = none (used to switch from the default Peskin interpolation stencil to the trilinear stencil).
<I>D3Q19</I> values = none (used to switch from the default D3Q15, 15 velocity lattice, to the D3Q19, 19 velocity lattice).
<I>read_restart</I> values = restart file = name of the restart file to use to restart a fluid run.
<I>write_restart</I> values = N = write a restart file every N MD timesteps.
<I>zwall_velocity</I> values = velocity_bottom velocity_top = velocities along the y-direction of the bottom and top walls (located at z=zmin and z=zmax).
<I>bodyforce</I> values = bodyforcex bodyforcey bodyforcez = the x,y and z components of a constant body force added to the fluid.
<I>printfluid</I> values = N = print the fluid density and velocity at each grid point every N timesteps.
</PRE>
</UL>
<P><B>Examples:</B>
</P>
<PRE>fix 1 all lb/fluid 1 2 1.0 1.0 setGamma 13.0 dx 4.0 dm 10.0 calcforce sphere1
fix 1 all lb/fluid 1 1 1.0 0.0009982071 setArea 1 1.144592082 dx 2.0 dm 0.3 trilinear noise 300.0 8979873
</PRE>
<P><B>Description:</B>
</P>
<P>Implement a lattice-Boltzmann fluid on a uniform mesh covering the LAMMPS
simulation domain. The MD particles described by <I>group-ID</I> apply a velocity
dependent force to the fluid.
</P>
<P>The lattice-Boltzmann algorithm solves for the fluid motion governed by
the Navier Stokes equations,
</P>
<CENTER><IMG SRC = "Eqs/fix_lb_fluid_navierstokes.jpg">
</CENTER>
<P>with,
</P>
<CENTER><IMG SRC = "Eqs/fix_lb_fluid_viscosity.jpg">
</CENTER>
<P>where rho is the fluid density, u is the local fluid velocity, sigma is
the stress tensor, F is a local external force, and eta and Lambda are the shear and
bulk viscosities respectively. Here, we have implemented
</P>
<CENTER><IMG SRC = "Eqs/fix_lb_fluid_stress.jpg">
</CENTER>
<P>with a_0 set to 1/3 (dx/dt)^2 by default.
</P>
<P>The algorithm involves tracking the time evolution of a set of partial
distribution functions which evolve according to a velocity discretized version
of the Boltzmann equation,
</P>
<CENTER><IMG SRC = "Eqs/fix_lb_fluid_boltzmann.jpg">
</CENTER>
<P>where the first term on the right hand side represents a single time relaxation
towards the equilibrium distribution function, and tau is a parameter physically
related to the viscosity. On a technical note, we have implemented a 15 velocity
model (D3Q15) as default; however, the user can switch to a 19 velocity model (D3Q19) through the use of the <I>D3Q19</I> keyword. This fix provides the user with the choice of two
algorithms to solve this equation, through the specification of the keyword <I>LBtype</I>. If
<I>LBtype</I> is set equal to 1, the standard finite difference LB integrator is used.
If <I>LBtype</I> is set equal to 2, the algorithm of <A HREF = "#Ollila">Ollila et al.</A> &#91;1&#93; is used.
</P>
<P>Physical variables are then defined in terms of moments of the distribution
functions,
</P>
<CENTER><IMG SRC = "Eqs/fix_lb_fluid_properties.jpg">
</CENTER>
<P>Full details of the lattice-Boltzmann algorithm used can be found in <A HREF = "#Mackay">Mackay et al.</A> &#91;2&#93;.
</P>
<P>The fluid is coupled to the MD particles described by <I>group-ID</I> through a
velocity dependent force. The contribution to the fluid force on a given lattice
mesh site j due to MD particle alpha is calculated as:
</P>
<CENTER><IMG SRC = "Eqs/fix_lb_fluid_fluidforce.jpg">
</CENTER>
<P>where v_n is the velocity of the MD particle, u_f is the fluid velocity
interpolated to the particle location, and gamma is the force coupling constant.
Zeta is a weight assigned to the grid point,
obtained by distributing the particle to the nearest lattice sites. For this,
the user has the choice between a trilinear stencil, which provides a support of
8 lattice sites, or the immersed boundary method Peskin stencil, which provides a
support of 64 lattice sites. While the Peskin stencil is seen to provide more
stable results, the trilinear stencil may be better suited for
simulation of objects close to walls, due to its smaller support. Therefore, by default, the Peskin stencil is used; however the user may switch to the trilinear stencil by specifying the keyword, <I>trilinear</I>.
</P>
<P>By default, the force coupling constant, gamma, is calculated according to
</P>
<CENTER><IMG SRC = "Eqs/fix_lb_fluid_gammadefault.jpg">
</CENTER>
<P>Here, m_v is the mass of the MD particle, m_u is a representative fluid mass at the particle location, and dt_collision is a collision time, chosen such that tau/dt_collision = 1 (see <A HREF = "#Mackay2">Mackay and Denniston</A> &#91;3&#93; for full details). In order to calculate m_u, the fluid density is interpolated to the MD particle location, and multiplied by a volume, node_area*dx_lb, where node_area represents the portion of the surface area of the composite object associated with a given MD particle. By default, node_area is set equal to dx_lb*dx_lb; however specific values for given atom types can be set using the <I>setArea</I> keyword.
</P>
<P>The user also has the option of specifying their own value for the force coupling constant, for all the MD particles associated with the fix, through the use of the <I>setGamma</I> keyword. This may be useful when modelling porous particles. See <A HREF = "#Mackay">Mackay et al.</A> &#91;2&#93; for a detailed description of the method by which the user can choose an appropriate gamma value.
</P>
<P>IMPORTANT NOTE: while this fix applies the force of the particles on the fluid,
it does not apply the force of the fluid to the particles.
When the force coupling constant is set using the default method, there is only one option to include this hydrodynamic force on the particles, and that is through the use of the <A HREF = "fix_lb_viscous.html">lb/viscous</A> fix. This fix adds the hydrodynamic force to the total force acting on the particles, after which any of the built-in LAMMPS integrators can be used to integrate the particle motion. However, if the user specifies their own value for the force coupling constant, as mentioned in <A HREF = "#Mackay">Mackay et al.</A> &#91;2&#93;, the built-in LAMMPS integrators may prove to be unstable. Therefore, we have included our own integrators <A HREF = "fix_lb_rigid_pc_sphere.html">fix lb/rigid/pc/sphere</A>, and <A HREF = "fix_lb_pc.html">fix lb/pc</A>, to solve for the particle motion in these cases. These integrators should not be used with the <A HREF = "fix_lb_viscous.html">lb/viscous</A> fix, as they add hydrodynamic forces to the particles directly. In addition, they can not be used if the force coupling constant has been set the default way.
</P>
<P>IMPORTANT NOTE: if the force coupling constant is set using the default method, and the <A HREF = "fix_lb_viscous.html">lb/viscous</A> fix is NOT used to add the hydrodynamic force to the total force acting on the particles, this physically corresponds to a situation in which an infinitely massive particle is moving through the fluid (since collisions between the particle and the fluid do not act to change the particle's velocity). Therefore, the user should set the mass of the particle to be significantly larger than the mass of the fluid at the particle location, in order to approximate an infinitely massive particle (see the dragforce test run for an example).
</P>
<HR>
<P>Inside the fix, parameters are scaled by the lattice-Boltzmann timestep, dt,
grid spacing, dx, and mass unit, dm. dt is set equal to (nevery*dt_MD), where dt_MD is the MD timestep.
By default, dm is set equal to 1.0, and dx is chosen so that tau/(dt) =
(3*eta*dt)/(rho*dx^2) is approximately equal to 1. However, the user has
the option of specifying their own values for dm, and dx,
by using the optional keywords <I>dm</I>, and <I>dx</I> respectively.
</P>
<P>IMPORTANT NOTE: Care must be taken when choosing both a value for dx, and a simulation domain size. This fix uses the same subdivision of the simulation domain among processors as
the main LAMMPS program. In order to uniformly cover the simulation domain with lattice sites,
the lengths of the individual LAMMPS subdomains must all be evenly divisible by dx.
If the simulation domain size is cubic, with equal lengths in all dimensions, and the default value for dx is used, this
will automatically be satisfied.
</P>
<P>Physical parameters describing the fluid are specified through <I>viscosity</I>,
<I>density</I>, and <I>a0</I>. If the force coupling constant is set the default way, the surface area associated with the MD particles is specified using the <I>setArea</I> keyword. If the user chooses to specify a value for the force coupling constant, this is set using the <I>setGamma</I> keyword.
These parameters should all be given in terms of the mass, distance, and time units
chosen for the main LAMMPS run, as they are scaled by the LB timestep,
lattice spacing, and mass unit, inside the fix.
</P>
<HR>
<P>The <I>setArea</I> keyword allows the user to associate a surface area with a given atom type. For example if a spherical composite object of radius R is represented as a spherical shell of N evenly distributed MD particles, all of the same type, the surface area per particle associated with that atom type should be set equal to 4*pi*R^2/N. This keyword should only be used if the force coupling constant, gamma, is set the default way.
</P>
<P>The <I>setGamma</I> keyword allows the user to specify their own value for the force coupling constant, gamma, instead of using the default value.
</P>
<P>The <I>scaleGamma</I> keyword should be used in conjunction with the <I>setGamma</I> keyword, when the user wishes to specify different gamma values for different atom types. This keyword allows the user to scale the <I>setGamma</I> gamma value by a factor, gammaFactor, for a given atom type.
</P>
<P>The <I>dx</I> keyword allows the user to specify a value for the LB grid spacing.
</P>
<P>The <I>dm</I> keyword allows the user to specify the LB mass unit.
</P>
<P>If the <I>a0</I> keyword is used, the value specified is used for the square of the speed of
sound in the fluid. If this keyword is not present, the speed of sound squared is
set equal to (1/3)*(dx/dt)^2. Setting a0 > (dx/dt)^2 is not allowed, as this may lead to instabilities.
</P>
<P>If the <I>noise</I> keyword is used, followed by a a positive temperature value, and a
positive integer random number seed, a thermal lattice-Boltzmann algorithm is used.
If <I>LBtype</I> is set equal to 1 (i.e. the standard LB integrator is chosen), the thermal LB algorithm of <A HREF = "#Adhikari">Adhikari et al.</A> &#91;4&#93; is used; however if <I>LBtype</I> is set equal to 2 both the LB integrator, and thermal LB algorithm described in <A HREF = "#Ollila">Ollila et al.</A> &#91;1&#93; are used.
</P>
<P>If the <I>calcforce</I> keyword is used, both the fluid force and torque
acting on the specified particle group are printed to the screen every N timesteps.
</P>
<P>If the keyword <I>trilinear</I> is used, the trilinear stencil is used to
interpolate the particle nodes onto the fluid mesh. By default, the
immersed boundary method, Peskin stencil is used. Both of these interpolation methods
are described in <A HREF = "#Mackay">Mackay et al.</A> &#91;2&#93;.
</P>
<P>If the keyword <I>D3Q19</I> is used, the 19 velocity (D3Q19) lattice is used by the
lattice-Boltzmann algorithm. By default, the 15 velocity (D3Q15) lattice is
used.
</P>
<P>If the keyword <I>write_restart</I> is used, followed by a positive integer, N,
a binary restart file is printed every N LB timesteps. This restart file
only contains information about the fluid. Therefore, a LAMMPS restart
file should also be written in order to print out full details of the
simulation.
</P>
<P>IMPORTANT NOTE: When a large number of lattice grid points are used, the
restart files may become quite large.
</P>
<P>In order to restart the fluid portion of the simulation, the keyword <I>read_restart</I>
is specified, followed by the name of the binary lb_fluid restart file to be used.
</P>
<P>If the <I>zwall_velocity</I> keyword is used y-velocities are assigned to
the lower and upper walls. This keyword requires the presence of walls
in the z-direction. This is set by assigning fixed boundary conditions
in the z-direction. If fixed boundary conditions are present in the
z-direction, and this keyword is not used, the walls are assumed to be stationary.
</P>
<P>If the <I>bodyforce</I> keyword is used, a constant body force is added to
the fluid, defined by it's x, y and z components.
</P>
<P>If the <I>printfluid</I> keyword is used, followed by a positive integer, N,
the fluid densities and velocities at each lattice site are printed to the
screen every N timesteps.
</P>
<HR>
<P>For further details, as well as descriptions and results of several test runs,
see <A HREF = "#Mackay">Mackay et al.</A> &#91;2&#93; . Please include a citation to this
paper if the lb_fluid fix is used in work contributing to published research.
</P>
<HR>
<P><B>Restart, fix_modify, output, run start/stop, minimize info:</B>
</P>
<P>Due to the large size of the fluid data, this fix writes it's own
binary restart files, if requested, independent of the main LAMMPS
<A HREF = "restart.html">binary restart files</A>; no information about <I>lb_fluid</I> is
written to the main LAMMPS <A HREF = "restart.html">binary restart files</A>.
</P>
<P>None of the <A HREF = "fix_modify.html">fix_modify</A> options
are relevant to this fix. No global or per-atom quantities are stored
by this fix for access by various <A HREF = "Section_howto.html#4_15">output
commands</A>. No parameter of this fix can be
used with the <I>start/stop</I> keywords of the <A HREF = "run.html">run</A> command.
This fix is not invoked during <A HREF = "minimize.html">energy minimization</A>.
</P>
<P><B>Restrictions:</B>
</P>
<P>This fix can only be used with an orthogonal simulation domain.
</P>
<P>Walls have only been implemented in the z-direction. Therefore, the
boundary conditions, as specified via the main LAMMPS boundary command
must be periodic for x and y, and either fixed or periodic for z.
Shrink-wrapped boundary conditions are not permitted with this fix.
</P>
<P>This fix must be used before any of <A HREF = "fix_lb_viscous.html">fix lb/viscous</A>, <A HREF = "fix_lb_momentum.html">fix lb/momentum</A>,
<A HREF = "fix_lb_rigid_pc_sphere.html">fix lb/rigid/pc/sphere</A>, and/ or <A HREF = "fix_lb_pc.html">fix lb/pc</A> , as
the fluid needs to be initialized before any of these routines try to access its
properties. In addition, in order for the hydrodynamic forces to be added to the particles, this fix must be used in conjunction with the <A HREF = "fix_lb_viscous.html">lb/viscous</A> fix if the force coupling constant is set by default, or either the <A HREF = "fix_lb_viscous.html">lb/viscous</A> fix or one of the <A HREF = "fix_lb_rigid_pc_sphere.html">lb/rigid/pc/sphere</A> or <A HREF = "fix_lb_pc.html">lb/pc</A> integrators, if the user chooses to specifiy their own value for the force coupling constant.
</P>
<P>This fix can only be used if LAMMPS was built with the
"fluid" package. See the <A HREF = "Section_start.html#2_3">Making
LAMMPS</A> section for more info on packages.
</P>
<P><B>Related commands:</B>
</P>
<P><A HREF = "fix_lb_viscous.html">fix lb/viscous</A>, <A HREF = "fix_lb_momentum.html">fix lb/momentum</A>, <A HREF = "fix_lb_rigid_pc_sphere.html">fix lb/rigid/pc/sphere</A>,
<A HREF = "fix_lb_pc.html">fix lb/pc</A>
</P>
<P><B>Default:</B>
</P>
<P>By default, the force coupling constant is set according to
</P>
<CENTER><IMG SRC = "Eqs/fix_lb_fluid_gammadefault.jpg">
</CENTER>
<P>and an area of dx_lb^2 per node, used to calculate the fluid mass at the particle node location, is assumed.
</P>
<P>dx is chosen such that tau/(delta t_LB) =
(3 eta dt_LB)/(rho dx_lb^2) is approximately equal to 1.
dm is set equal to 1.0.
a0 is set equal to (1/3)*(dx_lb/dt_lb)^2.
The Peskin stencil is used as the default interpolation method.
The D3Q15 lattice is used for the lattice-Boltzmann algorithm.
If walls are present, they are assumed to be stationary.
</P>
<HR>
<A NAME = "Ollila"></A>
<P><B>&#91;1&#93; (Ollila et al.)</B> Ollila, S.T.T., Denniston, C., Karttunen, M., and Ala-Nissila, T., Fluctuating lattice-Boltzmann model for complex fluids, J. Chem. Phys. 134 (2011) 064902.
</P>
<A NAME = "Mackay"></A>
<P><B>&#91;2&#93; (Mackay et al.)</B> Mackay, F. E., Ollila, S.T.T., and Denniston, C., Hydrodynamic Forces Implemented into LAMMPS through a lattice-Boltzmann fluid, Computer Physics Communications 184 (2013) 2021-2031.
</P>
<A NAME = "Mackay2"></A>
<P><B>&#91;3&#93; (Mackay and Denniston)</B> Mackay, F. E., and Denniston, C., Coupling MD particles to a lattice-Boltzmann fluid through the use of conservative forces, J. Comput. Phys. 237 (2013) 289-298.
</P>
<A NAME = "Adhikari"></A>
<P><B>&#91;4&#93; (Adhikari et al.)</B> Adhikari, R., Stratford, K., Cates, M. E., and Wagner, A. J., Fluctuating lattice Boltzmann, Europhys. Lett. 71 (2005) 473-479.
</P>
</HTML>
c4science · Help