Name | Modified | Size | Downloads / Week |
---|---|---|---|
Parent folder | |||
README | 2023-09-30 | 3.2 kB | |
scale_flow.f90 | 2023-09-30 | 9.0 kB | |
Totals: 2 Items | 12.3 kB | 0 |
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! ! program scale_flow ! ! Description: ! ! SCALE_FLOW is a companion in a limited sense to SCALE_GRID, which can ! apply numerous standard transformations to a multiblock structured grid's ! coordinates. This companion is limited to scaling or shifting one or more ! of the (flowfield?) functions in the input PLOT3D-type function file, in ! all blocks or in a subset (or subsets) of the blocks. [An additional ! option is described below.] ! ! Belated realization: Actually, ADJUST_GRID and ADJUST_FLOW are now the ! recommended pair for applying standard transformations. ! ! There is no need to read the accompanying grid file. The entire function ! file is read before a loop over scaling/shifting options is entered. ! The block dimensions and number of functions are not changed. ! ! Extending SCALE_GRID was considered but rejected because it uses the ! deprecated cfd_io_package. Handling of 2D grids is eschewed initially. ! ! Belated realization: ! ! Actually, ADJUST_GRID and ADJUST_FLOW are now the recommended pair for ! applying standard transformations. SCALE_GRID is deprecated. But ! SCALE_FLOW is retained as the name for this more specialized utility. ! ! Motivation: ! ! Difficulties with calculating an initial solution on an unaligned 3D whole ! body grid prompted SCALE_FLOW: conventional wisdom avoids revolving and ! reflecting an aligned 2D solution for fear of locking in some symmetric ! wake feature such as a disk shock that is a 2D artifact. Starting the 3D ! whole body solution from scratch can prove difficult or even impossible. ! This utility will allow revolving and reflecting the 2D state variables ! then artificially introducing some asymmetry in one or more of those state ! variables, probably in just the aft blocks, in the hope of providing a ! decent starting guess with a fully developed wake that will not preserve ! any of the initial 2D artifacts. ! ! Another option is included in the hope of assisting calculations from ! scratch (essentially from free-stream conditions everywhere): these ! solutions can stall out (NaN) with the wake only partially developed. ! Copying k planes from the partially-developed wake into the remaining ! downstream k planes may be beneficial (for DPLR-type grids with a single ! layer of blocks). ! ! Control: ! ! A few prompts serve to control this utility. ! ! History: ! ! 09/19/2023 D.A.Saunders Initial design. ! 09/20/2023 " " Implementation and testing. ! 09/23/2023 " " Reading of large function files can be very slow ! because of %q(m,i,j,k) indexing of xyzq_io. We ! really need a %q(i,j,k,m) option. A bit more ! run-time dialogue was desirable. ! 09/25/2023 " " A %q(i,j,k,m) option is now available and used. ! ! Author: David Saunders, AMA, Inc. at NASA Ames Research Center, CA. ! !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!