Research Group of Prof. Dr. M. Griebel
Institute for Numerical Simulation
next up previous contents
Next: Objects Up: Scene description file Previous: dimension   Contents


The parameter-block contains several parameters which control the computation and several constants describing the flow and physical quantities. If a parameter is not given, a default value will be used. A complete list of all possible parameters together with an explanation and the default value is given in table [*].

All physical quantities, like Dirichlet values for the velocity, temperature or the viscosity are assumed to be given in standard ISO units (m, s, kg, K and derived units), unless the dimensionless-flag is set(see below).

Table: List of entries in the parameter-block
Name Type Default Description
Flow parameters
Tfin double [s] 1.0 defines the (physical) timespan of the simulation
reynolds double 10.0 is the dimensionless Reynolds-number which describes the ratio between inertia and viscous forces in the flow
nuC int 0 the paramter nuC sets the number of species to be transported with the flow. For each species, a diffusion constant has to be specified in a comma-separated list after nuC, i.e. the line has to look like nuC $ n_C, \nu_1, \dots, \nu_{n_C}$ where $ n_C$ is an integer specifying the number of species and $ \nu_i$ is a double value which stands for the diffusion parameter of the i-th species. You have to make sure that you also specify initial conditions using the keyword cheminit. Details on how to specify initial and boundary conditions are given in section [*]
gx double [m/s$ ^2$] 0.0 $ x$-component of external volume force $ {\bf g}$ in ([*]) (or $ {\bf g}_0$ in ([*]) if the temperature is calculated). In the latter case $ {\bf g}$ is computed from $ {\bf g}_0$ by means of ([*])
gy double [m/s$ ^2$] 0.0 Same as gx for the $ y$-component
gz double [m/s$ ^2$] 0.0 Same as gx for the $ z$-component
froude double 1.0 the Froude-number is a dimensionless number describing the ratio between inertial and gravitational forces
beta double [1/K] 1e-4 volume expansion coefficient in equation ([*])
TempRef double [K] 273.0 reference temperature in ([*])
prandtl double 1.0 the Prandtl-number is a dimensionless number describing the ratio between momentum and heat transfer in the fluid
Timestep control
deltmax double [s] 1.0 upper bound for $ \Delta t$
tfconv double 0.1 security factor for the timestep restriction arising from the convective terms. In every step, $ \Delta t$ will be set less or equal than

$\displaystyle \Delta t\leq{\tt tfconv}\cdot\min_{c\in{\cal F}}\Big\{\frac{\Delta x_c}{u_c},
\frac{\Delta y_c}{v_c},\frac{\Delta z_c}{w_c}\Big\} $

where $ {\cal F}$ denotes the set of cells

$\displaystyle c=[x,x+\Delta x_c]\times [y,y+\Delta y_c]\times [z,z+\Delta z_c] $

in $ \Omega$
tfdiff double 0.2 security factor for the timestep-restriction arising from the diffusive terms. In every step, $ \Delta t$ will be set less or equal than

$\displaystyle \Delta t\leq{\tt tfdiff}\min_{\nu\in{\cal N},c\in{\cal F}}
...1{{\Delta x_c}^2}+\frac1{{\Delta y_c}^2}+\frac1{{\Delta z_c}^2}\Big)\Big]^{-1} $

and $ {\cal N}=\{\nu\}\cup\underbrace{\{1/(\nu\cdot{\tt prandtl})\}}_
\mbox{\small only if scalars}\\
\mbox{\small are computed}

Data output
prstep int 20 defines when to write the computed values to the binary file. Each prstep-th timestep the current solution is written to the binary file(the file is overwritten)
TimePrintStep string - defines an interval (in physical time) after which the solution should be written to files in a directory specified by TargetDirectory. If TimePrintStep is not specified, no output will be generated.
TargetDirectory string - specifies the directory(absolut pathname) where the files generated by TimePrintStep should be stored
Parameters controlling numerical methods
TimeDis string EU1 defines the time discretization to be used. Possible values are EU1 for first order explicit Euler-Method, AB2 for second order explicit Adams-Bashforth-Method, RK2 for second order Runge-Kutta-method and RK3 for third order Runge-Kutta-method. Remark: The Runge-Kutta-method of third order is only available for the time derivatives in equations ([*]) and ([*]), setting TimeDis to RK3 results in application of the Runge-Kutta-method of second order to the other time derivatives
ConvectiveTerms string VONOS defines the discretization scheme to be used for the convective terms. Possible values are DC (Donor-Cell, 1st/2nd order), HLPA (Hybrid Linear-Parabolic Approximation, 1st/2nd order), QUICK (Quadratic Upwind Interpolation for Convective Kinematics, 2nd order), SMART (Sharp And Monotonic Algorithm for Realistic Transport, 2nd order) and VONOS (Variable-Order Non-Oscillatory Scheme, 2nd order)
PoissonSolver string BiCGStab set the method for solution of the linear system arising from discretization of the pressure poisson equation. Possible values are SOR, SSOR, RedBlack, 8ColorSOR, 8ColorSSOR and BiCGStab (preconditioned with Jacobi-Method)
alpha double 1.0 defines the blending parameter $ \alpha$ in the convex combination of the central difference/upwind discretization of the convective terms of $ F,G,H$. $ {\tt alpha}=1$ means pure upwind and $ {\tt alpha}=0$ results in pure central difference discretization
alphaTC double 1.0 same as alpha, but for convective terms in the transport equation used for computation of temperature and scalars
Parameters for the linear solver
itermax int 100 defines the maximal number of iterations in the linear solver (BiCGStab,SOR, SSOR etc.)
eps double 0.001 defines the stopping criterion for the iterations in the linear solver. The parameter eps is the upper bound for the residual of the poisson equation, i.e. the iterations are stopped if

$\displaystyle \sqrt{\frac{1}{\sharp{\cal F}}
\sum_{c\in{\cal F}}\Big(\Delta_hp_c -\frac{1}{\Delta t} (\nabla_h\cdot \tilde{\bf u})_c\Big)^2}\leq\mbox{\tt eps}~.

Here, $ {\cal F}$ is the set of all fluid cells in $ \Omega$ and $ \sharp{\cal F}$ is the cardinality of $ {\cal F}$. $ \Delta_h$ and $ \nabla_h$ denote the discrete Laplacian and gradient operator.
omega double 1.7 sets the relaxation parameter for the SOR-type solvers
Boundary conditions
periodboundx     sets periodic boundary conditions in direction of $ x$-coordinate
periodboundy     same as periodboundx only for the $ y$-coordinate-direction
periodboundz     same as periodboundx only for the $ z$-coordinate-direction


next up previous contents
Next: Objects Up: Scene description file Previous: dimension   Contents
Martin Engel 2004-03-15