ACCELERATE ZONE range1[,range2[, ... rangen]] [CFLRAMP START iter_{1} END iter_{2}] [SECOND START iter_{1} END iter_{2}] [FOURTH START iter_{1} END iter_{2}] ENDACCELERATE |
ACCELERATE ZONE range1[,range2[, ... rangen]] [CFLRAMP START iter_{1} END iter_{2}] ENDACCELERATE |
The objective of Wind-US's convergence acceleration scheme is to approach the steady state solution more quickly by allowing the use of a large CFL number early in the calculation.
The ZONE keyword identifies the zone(s) within which convergence acceleration is to be used. The range parameter(s) must be one of the following forms:
zonenum | Selects zone zonenum | ||
---|---|---|---|
begzone:endzone | Selects all zones from begzone to endzone | ||
ALL | Selects all zones |
The CFL number will be ramped down linearly from its starting value, specified using the CFL# keyword, to 1.0. The ACCELERATE keyword does not work when CFL# MODE 3 (i.e., the time step calculation procedure from OVERFLOW) is used.
With structured grids, first- and second-order smoothing are normally applied to maintain numerical stability. The amount of smoothing, when it is applied, and how it varies with iteration is controlled by the SMOOTHING and/or BOUNDARY-DAMP keywords, the iter_{1} and iter_{2} values specified with ACCELERATE, and the mode used with TEST option 49.
For the second-order smoothing,
mode | Second-Order Smoothing
| |||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
< 2 | Smoothing varies with iteration i, as follows:
| |||||||||||
2 | Smoothing varies with iteration i, as follows:
| |||||||||||
3 | No smoothing applied |
mode | Fourth-Order Smoothing
| ||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
< 2 | Smoothing applied as determined by the SMOOTHING keyword
| ||||||||||||||
2 | Smoothing varies with iteration i, as follows:
| ||||||||||||||
3 | Smoothing applied as determined by the SMOOTHING and/or BOUNDARY-DAMP keywords |
If iter_{1} and iter_{2} aren't specified,
the following default values are used.
Option | iter_{1} | iter_{2} | |||
---|---|---|---|---|---|
SECOND | 1 | 150 | |||
FOURTH | 50 | 500 | |||
CFLRAMP | 50 | 500 |
Recommended values for the second- and fourth-order smoothing parameters
val_{1} and val_{2} specified with the
SMOOTHING keyword, and for the
starting CFL number, are:
Dimensions | val_{1} | val_{2} | CFL # | ||||
---|---|---|---|---|---|---|---|
2D | 0.1 | 0.03 | 10-15 | ||||
3D | 0.06 | 0.01 | 2.5 |
In the following example, convergence acceleration is applied in zones 3 and 4. Second-order smoothing is applied between iterations 1 and 200, and fourth-order smoothing is applied between the default values of iterations 50 and 500. The CFL number specified with the CFL# keyword will be used for the first 250 iterations, then ramped down to a value of 1.0 between iterations 250 and 500.
SMOOTHING SECOND 0.06 FOURTH 0.01 SMLIMT 0. ACCELERATE ZONE 3,4 SECOND START 1 END 200 CFLRAMP START 250 END 500 ENDACCELERATE TEST 49 2
See Also: SMOOTHING, BOUNDARY-DAMP, CFL#, TEST 49
The ZONE keyword identifies the zone(s) within which convergence acceleration is to be used. The range parameter(s) must be one of the following forms:
zonenum | Selects zone zonenum | ||
---|---|---|---|
begzone:endzone | Selects all zones from begzone to endzone | ||
ALL | Selects all zones |
iter_{1} and iter_{2} specify starting and ending iteration values for varying the CFL number. Just as for structured grids, for unstructured grids the CFL number will be ramped down linearly from its starting value, specified using the CFL# keyword, to 1.0. If the CFLRAMP keyword isn't specified, default values of 50 and 500 are used, for iter_{1} and iter_{2}.
For unstructured grids, however, no variation in smoothing is done.
See Also: CFL#
Last updated 17 Feb 2009