CFD Utilities Files

Prepare_local_analysis Description:

   This is a variant of prepare_rapid_analysis with an extra prompt for the
   number of species in the flow solution.  Renaming it avoids affecting
   existing automation scripts. This variant was prompted by turbulent flow
   cases, where it is no longer feasible to handle 1 or 2 temperatures in
   air without the prompt that affects existing automation.

Site-Specifics:

   If this program is installed at another site, the following path should
   be edited appropriately prior to compiling and linking:

      /share/apps/cfdtools/         ! See parameter constant "path" below

   It is assumed that all the utilities invoked here are in subdirectories
   with uppercase names and this indicated path.

   E.g., thin_grid, adjust_grid, ... should be installed here:

      /share/apps/cfdtools/THIN_GRID/thin_grid
      /share/apps/cfdtools/ADJUST_GRID/adjust_grid   and so on.

   The following utilities may be invoked by the script from this program:

      thin_grid           ! Option to thin new input surface/cavity grids
      adjust_grid         ! Option to convert input inches to meters
      combine_blocks_turb ! Option to append/init. cavity blocks (lam|turb)
      template            ! Generates dplr interface & control files
      plug_interp         ! Specialized treatment of a wing LE plug

Description (mostly from the original prepare_rapid_analysis):

   Program prepare_local_analysis generates a shell script, named
   local_analysis_script, and control files to run the NASA Ames tools for
   rapid CFD analysis of damage configurations (or other perturbations -
   often a tile cavity) with minimal user interaction.  The intended flow
   solver is DPLR.

   The baseline grid is assumed to have a single layer of blocks, with
   k = 1 at the geometry surface.

   Another shell script may be used to run prepare_local_analysis, to make
   its output script executable, and to invoke that script. At NASA ARC see

      /share/apps/cfdtools/PREPARE_LOCAL_ANALYSIS/local_analysis.sh

   which is also pointed to by a symbolic link that should be in your path:

      /share/apps/cfdtools/bin/local_analysis

   Then typing "local_analysis" performs everything between the Gridgen and
   FCONVERT steps via answers to a few prompts.

   Alternatively, run prepare_local_analysis directly, then use

      source local_analysis_script

   to execute its output script.

Usual sequence of steps:

   Given a baseline grid (vertex-centered) and a flow solution with the 
   cell-centered form of the grid, including "halo" cells, plus a new
   surface grid (probably confined to the locality of a surface perturb-
   ation) and (commonly) an associated cavity grid block or plug/gap
   filler collar blocks, the usual sequence is:

   1:  Thin the new surface and volume grids (THIN_GRID with 2 2 1 inputs).

   2:  Convert the thinned grids from inches to meters (ADJUST_GRID).

   3:  Generate a volume grid from the new surface & baseline volume grids
       (first application of RADIAL_INTERP, with vertex-centered data).

   4:  Interpolate the flow soln. to this volume grid (second application
       of RADIAL_INTERP, with cell-centered data but same new surface grid
       as 3).

   5:  Combine the main volume grid from 3 with the ancillary block(s)
       (first application of COMBINE_BLOCKS, with vertex-centered data).

   6:  Combine the main flow starting guess with estimates for ancillary
       block(s) calculated as a specialized option by a second application
       of COMBINE_BLOCKS (cell-centered data except for the cavity grid).

   7:  Generate DPLR control files (dplr.inputs and dplr.interfaces) using
       TEMPLATE.  A sample.inputs file should be present so that the header
       and trailer records can be transcribed to dplr.inputs.  Some editing
       of dplr.inputs may still be needed:  the correct number of blocks,
       the correct flight conditions, and the appropriate CFL schedule.

N.B.:  Note that TEMPLATE can now avoid the need to edit certain BCs for
       the cavity blocks (and different BCs for blocks around a plug repair
       or a protruding gap filler) via the 'template.inp.2' ancillary file.
       Ideally, the present program would write such a file based on how
       many blocks are in the (optional) second grid file.  Instead, if the
       file is present, this program can use it to distinguish between a
       cavity and a plug/gap filler case.  Otherwise, it assumes it is
       neither - probably just a new surface grid.

       Note that gap filler cases are treated as for plug cases, and should
       contain 'plug' in template.inp.2 (case-insensitive).
   
   Two runs of FCONVERT could in principle be added, but how to split the
   blocks may need to be decided at run time according to how many CPUs are
   deemed appropriate.

History:

   11/28/05  DAS  Initial step towards automating rapid cavity analysis.
   11/29/05   "   Added the THIN_GRID, SCALE_GRID and TEMPLATE options.
   01/09/06   "   Ryan McDaniel ran into a glitch in the no-cavity case,
                  and suggested a prologue to summarize requirements.
   03/31/06   "   Remind the user about editing the cavity wall BCs, and
                  default the baseline volume grid and solution files.
   04/14/06   "   Added handling of the leading edge plug repair case
                  and exploited the new TEMPLATE option to make use of
                  a 'template.inp.2' control file.
   04/17/06   "   Fixed a couple of typos in the prologue.
   08/04/06   "   RADIAL_INTERP's distance tolerance should be negative,
                  meaning it will decay any surface discrepancies to
                  zero at the outer boundary.
   10/24/06   "   Changed the default Shuttle baseline file names from
                  "right half" to "both halves".  Documented the fact
                  that protruding gap filler cases are treated as for
                  plug cases at this level.
   07/11/08   "   The output script now begins with #! /bin/tcsh.
   10/15/08   "   It now removes any RADIAL_INTERP log files that might
                  be present from an earlier run.
   10/19/08   "   A Todd White case broke the assumption that the new
                  surface and volume files are formatted the same way.
                  That is no longer assumed.
   09/09/09   "   PREPARE_LOCAL_ANALYSIS is just PREPARE_RAPID_ANALYSIS
                  with one extra prompt for cavity-type cases (number of
                  species in the flow solution).  Renaming it avoids
                  impacting automation scripts.
   09/11/09   "   The prologue shouldn't mention RADIAL_INTERP control
                  files because these are generated here once all the
                  prompts have been answered.
   02/13/10   "   Decaying surface discrepancies to zero at the outer
                  boundary is found to be the wrong choice for a WLE
                  plug (working with an offset surface in a high-
                  curvature region).  Only shallow cavities really
                  need the decay, so suppress it here via the +0.0002
                  inputs to RADIAL_INTERP that had been -0.0002.
   01/18/12   "   Installation on Redwood-2 prompted use of a path to
                  the utilities.  But all of them have uppercase
                  directories now, instead of the earlier mixture.
   07/19/12   "   An extra prompt allows the decay/no-decay choice now.
   01/13/14   "   ADJUST_GRID is preferable to SCALE_GRID now, so the
                  option to convert inches to meters has been recoded.
   04/19/17   "   Dinesh Prabhu encountered a glitch with cavity blocks
                  that should have been encountered long ago: gridname
                  was used instead of nspecies in a READER prompt.
   06/16/17   "   Dinesh suffered from another glitch: the pscale prompt for
                  COMBINE_BLOCKS_TURB was missing, and the cavity flow was
                  coming out all zeros for the species densities.  It turns
                  out that an extra 'n' for the format of the combined flow
                  file (written here) was being read by COMBINE_BLOCKS_TURB
                  as pscale and coming out zero.  That 'n' should not be
                  there because the formatting of the combined grid is used.
   
Author:  David Saunders, ELORET Corp./NASA Ames Research Center, CA
                         Now ERC, Inc./NASA ARC.