TRACE User Guide
TRACE Version 9.6.1
|
Activate settings for viscous timestep calculations. Include spectral radii of the viscous flux Jacobians in the computation of timesteps.
viscCFL | viscous CFL number |
Enable the use of the legacy compatibility mode. This mode will yields a solver behaviour of spliss which is equivalent to the legacy linear solver. Spliss per default uses a coloring algorithm which favors thread parallelism, but is weaker in terms of algorithmic performance. Note that this mode does NOT re-enable the legacy solver if TRACE was compiled with USE_SPLISS. Also note, that this mode only affects unstructured blocks. Structured blocks are already equivalent to the legacy linear solver anyway.
ON | enable use of the legacy compatibility mode |
OFF | disable use of the legacy compatibility mode |
Differentiate between implicit and explicit nodes at the cell level (rather than just at the block level).
If convergence control is active, hinder convergence if a CFL penalty is active.
Start to compute temporal error estimated by IRK and writes error into 3D CGNS file.
Stop the computation of the temporal error estimated by IRK and writing the error into 3D CGNS file.
Force the use of the implicit part of the IMEX time-integration method in the specified block. Can only be used in combination with IMEX Runge-Kutta methods.
Enable the use of ILU(0) diagonal entries if the ILU solver is used for the implicit system in flow and transport equations. For the SGS and SSOR solver this command has no effect.
ON | enable use of ILU(0) diagonal entries |
OFF | disable use of ILU(0) diagonal entries (default) |
Sets the number of pseudo time iterations used in the dual-time stepping algorithm every physical time step.
number | Number of pseudo-time iterations. |
stage | Stage of IRK scheme. (if IRK is active) |
Activate a particular time-integration method.
method | (one of CrankNicolson , EulerBackward , ImplicitRungeKutta , ExplicitRungeKutta , ImplicitExplicitRungeKutta ). |
Activate a (non-default) Runge-Kutta method.
method | (one of FourthOrderExplicitFiveStage , SecondOrderExplicitTwoStage , ThirdOrderExplicitFourStage ). |
Set integration method.
method | Integration method (one of PredictorCorrector , GaussSeidel ). |
Change time integration accuracy.
order | accuracy (one of 1stOrder , 2ndOrder for BDF; one of 1stOrder , 2ndOrder , 3rdOrder , 4thOrder , 5thOrder for IRK). |
Set the maximum absolute time step at which TRACE terminates. TRACE may terminate earlier if convergence/divergence criteria are met.
number_of_time_steps | absolute time steps of the simulation (integer) |
Set the maximum - relative to the initial time step - number of iterations till TRACE terminates (initial time step + iterations = number of total time steps). TRACE may terminate earlier if convergence/divergence criteria are met.
number_of_iterations | iterations till termination of the simulation (integer) |
Set global time step size.
dt | time step size in seconds |
Set the number of time-steps over which the solution is to be averaged in order to compute time-averaged values.
nTimeStepsPerPeriod | number of time-steps required to integrate over a single period |
Set parameters for CFL ramp.
startCFLvalue | CFL number at start of ramp. |
targetCFLvalue | CFL number at end of ramp. |
duration | Number of time steps from current to reach target. |
Sets the CFL number of the selected blocks or activates automatic CFL number control via the AUTO keyword.
The automatic CFL number control evaluates the slope of one of the Monitor functions:
L1_global
global L1 residual of the flow equationsL2_global
global L2 residual of the flow equationsLmax_global
global maximum residual of the flow equationsL1_family
block-family L1 residual of the flow equationsL2_family
block-family L2 residual of the flow equationsLmax_family
block-family maximum residual of the flow equationsL1_block
block-local L1 residual of the flow equationsL2_block
block-local L2 residual of the flow equationsLmax_block
block-local maximum residual of the flow equationsIn case of a negative slope the non-penalized CFL number is increased by the specified GrowthRate, otherwise it is decreased by the specified DecayRate. Furthermore, it is limited to the specified LowerBound and UpperBound.
CFL|AUTO | CFL number or keyword AUTO for automatic CFL number control with additional parameters |
Monitor | one of the above monitor functions used for CFL number control |
DecayRate | CFL decay rate in case of a positive or constant slope of the chosen monitor function (0 < DecayRate < 1) |
GrowthRate | CFL growth rate in case of a negative slope of the chosen monitor function (0 < GrowthRate < DecayRate / (1 - DecayRate)) |
LowerBound | lower bound for the non-penalized CFL number (0 < LowerBound) |
UpperBound | upper bound for the non-penalized CFL number (LowerBound <= UpperBound) |
Configures CFL penalization of blocks that encounter a critical flow state. In that case the block is reset and its CFL number is penalized by the PenaltyFactor. The CFL penalty recovers in the following iteration steps by the PenaltyRecoveryRate up to the non-penalized CFL number. If the penalized CFL number drops below the MinimumAllowedCflNumber the simulation aborts. In robust mode only a warning is issued. If the penalized CFL number reaches a value of zero the simulation aborts unconditionally.
PenaltyFactor | penalization per failed solution update (0 < PenaltyFactor < 1, default: 0.9) |
PenaltyRecoveryRate | recovery rate of penalization per successful solution update (0 < PenaltyRecoverRate, default: 0.1) |
MinimumAllowedCflNumber | minimum penalized CFL number that is tolerated before the simulation aborts (inactive if MinimumAllowedCflNumber <= 0, default: 1e-6) |
OFF | deactivates the CFL penalty for steady and unsteady time-domain |
In unsteady time-domain, TRACE throws a serious if the solution update fails in all sub-iterations. With command TolerateNoSolutionUpdate, a warning is issued instead of a serious. It is tolerated that the flow solution is not updated during a time step, i.e. active CFL penalty in all sub-iterations. Command can only be used for blockgroups in unsteady mode, otherwise the simulation aborts.
Sets the pseudo-time stepping mode of the pseudo-time solvers.
local | cell based local pseudo time stepping (pseudo-time step computed separately for each harmonic) (default) |
localAllHarmonics | cell based local pseudo time stepping (pseudo-time step computed identically for all harmonics, in time-domain solver modes equivalent to local) |
block | block based pseudo time stepping (pseudo-time step computed separately for each harmonic) |
blockAllHarmonics | block based pseudo time stepping (pseudo-time step computed identically for all harmonics, in time-domain solver modes equivalent to block) |
family | family based pseudo time stepping (pseudo-time step computed separately for each harmonic) |
familyAllHarmonics | family based pseudo time stepping (pseudo-time step computed identically for all harmonics, in time-domain solver modes equivalent to family) |
global | global pseudo time stepping means same time step in the entire domain (pseudo-time step computed identically for all harmonics) |