PCGOPT, Lev_Diff
, --
, ReduceIO
, StrmCk
, Wrtfull
, Memory
, LM_Key
Controls PCG solver options.
Lev_Diff
Indicates the level of difficulty of the analysis.
Valid settings are AUTO or 0 (default), 1, 2, 3, 4, or 5. This option
applies to both the PCG solver when used in static and full transient
analyses and to the PCG Lanczos method in modal analyses. Use AUTO
to let ANSYS automatically choose the proper level of difficulty for
the model. Lower values (1 or 2) generally provide the best performance
for well-conditioned problems. Values of 3 or 4 generally provide
the best performance for ill-conditioned problems; however, higher
values may increase the solution time for well-conditioned problems.
Higher level-of-difficulty values typically require more memory. Using
the highest value of 5 essentially performs a factorization of the
global matrix (similar to the sparse solver) and may require a very
large amount of memory. If necessary, use Memory
to reduce the memory usage when using Lev_Diff
= 5. Lev_Diff
= 5 is generally recommended
for small- to medium-sized problems when using the PCG Lanczos mode
extraction method.
For example, models containing elongated elements (i.e., elements with high aspect ratios) and models containing contact elements can lead to ill-conditioned problems. To determine if your problem is ill-conditioned, view the Jobname.PCS file to see the number of PCG iterations needed to reach a converged solution. Generally, static or full transient solutions that require more than 1500 iterations are considered to be ill-conditioned for the PCG solver.
--
Unused field.
ReduceIO
Controls whether the PCG solver will attempt to reduce I/O performed during equation solution:
AUTO | — | Automatically chooses whether to reduce I/O or not (default). |
YES | — | Reduces I/O performed during equation solution in order to reduce total solver time. |
NO | — | Does NOT reduce I/O performed during equation solution. |
This option applies to both the PCG solver when used in static and full transient analyses and to the PCG Lanczos method in modal analyses.
StrmCk
Controls whether or not a Sturm sequence check is performed:
OFF | — | Does NOT perform Sturm sequence check (default). |
ON | — | Performs Sturm sequence check |
This option applies only when using the PCG Lanczos method in modal analyses. When using this option, a factorization must be performed and will require a very large amount of memory for extra computations. This option is generally recommended for small- to medium-sized problems. If the Sturm sequence check takes a large amount of computing time, use the Jobname.ABT file to abort the Sturm check, or press the button if in interactive mode.
Wrtfull
Controls whether or not the .FULL file is written.
ON | — | Write .FULL file (default) |
OFF | — | Do not write .FULL file. |
This option applies only when using the PCG Lanczos method in modal analyses because the .FULL file is never written when using the PCG solver in static or full transient analyses.
If using MSAVE,ON and conditions for the MSAVE command are met, a complete .FULL file is never written regardless of this option.
If constraint equations are present in the model, a .FULL file is always written regardless of this option.
This option is useful in Distributed ANSYS because assembling the global
stiffness and mass matrices on the master machine before writing the .FULL file can take a considerable amount of memory. By
setting Wrtfull
= OFF, this assembly process
is skipped on the master machine, decreasing the amount of memory
required to compute the modes and mode shapes. Wrtfull
= OFF does not affect the results for the modes and mode shapes.
However, without a .FULL file, the participation
factor table computations do not occur.
To generate the .FULL file, such as for
a harmonic, transient mode-superposition, or spectrum analysis, rerun
the modal analysis with Wrtfull
= ON, or
use the WRFULL command.
Memory
Controls whether to run using in-core or out-of-core
mode when using Lev_Diff
= 5.
AUTO | — | Automatically chooses which mode to use (default). |
INCORE | — | Run using in-core mode. |
OOC | — | Run using out-of-core mode. |
This option applies to both the PCG solver when used in static
and full transient analyses and to the PCG Lanczos method in modal
analyses. However, this option only applies when using Lev_Diff
= 5 and does not apply to Distributed ANSYS. Forcing
the INCORE option will require a large amount of memory and is only
recommended for small-to-medium sized problems. Using the OOC option
will make the PCG solver use the least amount of memory (relative
only to Lev_Diff
= 5) at the cost of additional
I/O, which will likely decrease performance as compared to an in-core
run.
LM_Key
Controls use of the PCG solver for MPC184 Lagrange multiplier method elements. This option applies only to the PCG solver when used in static and full transient analyses.
The Lagrange multiplier method used by MPC184 elements transfers into multiple point constraints
and, hence, can be solved by the PCG solver. For Lagrange-formulation
contact methods, incompressible u-P formulations, and the MPC184 screw joint element, the PCG solver cannot be used and the sparse solver is required. Distributed ANSYS does
not support the LM_Key
= ON option. The
MSAVE command does not support the LM_Key
= ON option.
ReduceIO
works independently of the MSAVE command in the PCG solver. Setting ReduceIO
to YES can significantly increase the memory
usage in the PCG solver.
To minimize the memory used by the PCG solver with respect to
the Lev_Diff
option only, set Lev_Diff
= 1 if you do not have sufficient memory
to run the PCG solver with Lev_Diff
=
AUTO.
The MSAVE,ON command is not valid when using Lev_Diff
= 5. In this case, the Lev_Diff
value will automatically be reset to 2. The MSAVE,ON command is also not valid with the StrmCk
option. In this case, StrmCk
will be
set to OFF.
Distributed ANSYS Restriction The Memory
option and the LM_Key
option are not supported in Distributed ANSYS.