Program NEQAIR_DATA Description:
Version Description:
This version of NEQAIR_DATA can write line-of-sight (LOS) data in both the
original rigid format read by NEQAIR versions up to and including v14 and the
simpler column-oriented format handled from NEQAIR v15 on. Both data formats
involve a sample.LOS.dat file read here that indicates the species names. If
a sample file is found with a NEQAIR version number on line 1, then the later
data format is indicated. The sample file is transcribed as the header of all
output LOS data files in either case.
Original Preamble:
Earlier utilities LINES_OF_SIGHT[_2D] and FLOW_INTERP[_2D] typically
produce multiple lines of sight and flow-field data interpolated to each
point of each line of sight, as a PLOT3D-type multiblock grid and function
file pair. Note that the input coordinates should correspond to grid
vertices, not cell centers, because the lines of sight are discretized using
the relative distribution of the nearest radial line from the vertex-
centered grid.
Radiation program NEQAIR treats one line of sight at a time, and expects
the data in a rigid format containing arc lengths (measured in the direction
towards the body, and in cm), total number density, T, Tr, Tv, Te, and the
species number densities, at each grid point along the line. The arc lengths
should correspond to the centers of cells at which the flow function data
are taken to apply (in NEQAIR), so they are adjusted accordingly here. The
wall function values thus apply to a cell half the width of the grid cell
at the wall.
[Later: This version has the option, prompted by meteor studies, to
output line-of-sight data in the direction away from the body.]
This program retrieves the PLOT3D-format files (e.g., los.g & los.f in
SI units) and writes a separate file in NEQAIR's LOS.dat format for each
line of sight it finds, into the working directory. These output file names
contain line numbers, such as LOS-1.dat. A script can then run NEQAIR on
some range m:n of line numbers, by copying or linking to input files and
writing radiation results for each line one level down in directories
/LINE-m through /LINE-n (say).
The current version now offers options to redistribute the line-of-sight
data to a [presumably smaller] different number of points in order to speed
the expensive NEQAIR calculations. This was initially done based on |dT/ds|,
where T is vibrational temperature and s is arc length along the line. (Use
of the translational temperature, which is higher than Tv at the shock, tends
to produce spacing that is too coarse in the boundary layer.) See below for
further redistribution options.
Method:
The grid and function file may be 2D or 3D (automatically determined).
The input grid coordinates are assumed to be in meters, emanating from the
body towards the shock. Normal usage with the DPLR flow solver has block
dimensions (1,nj) or (1,1,nk), but these can safely be treated here as (nj,1)
or (nk,1,1) for more efficient Fortran indexing. Points outside the shock
need to be suppressed to avoid NEQAIR trouble with low temperatures, and the
line data need to be output in the shock-to-body direction.
The following paragraph is NEQAIR V14-specific:
The function file should contain T, Tr, Tv, Te (K) and the species number
densities (m^-3). Note that DPLR's postprocessor does NOT extract Tr unless
irot = 1. BE SURE THAT THERE ARE 4 TEMPERATURES IN THE FUNCTION FILE.
Common usage with 2-temperature data (ivib = 1 in DPLR) means reading and
writing T, T, Tv, Tv for these 4 temperatures.
NEQAIR V15 differences:
NEQAIR V15[+] handles variables numbers of temperatures. The number of
species names beginning with T determines the number of temperatures that
should be in the input function file in the indicated order.
One line of sight's worth of data (one input block) is read at a time.
The header of a file named sample.LOS.dat file in the working directory is
transcribed to each output file, down to the line above the first line of
the data proper for line 1. This sample file should contain the species
names in the order matching the input function file. This program is thus
independent of the number and order of the species, as long as the file
sample.LOS.dat lists them in the header in the correct order.
Minimum Contents of File sample.LOS.dat:
V14 or earlier:
> A line of asterisks precedes the species section.
> The species names may appear in any order.
> Two lines of hyphens must appear. The last line is all hyphens.
> Replace the comments in column 1 below appropriately.
*****************************************************************************
aaaaaaaa aaaaaaaa aaaaaaaa aaaaaaaa (2x,(7x,a8))
CO2 CO N2 O2 :Species Symbols.
NO C N O
-----------------------------------------------------------------------------
no. x,cm total partcc t tr tv te (i5,f8.3,
ii rrrrrrr rrrrrrrrrrrrrr rrrrrrrrr rrrrrrrrr rrrrrrrrr rrrrrrrrre15.6,4f10.1
rrrrrrrrrrrrrr rrrrrrrrrrrrrr rrrrrrrrrrrrrr rrrrrrrrrrrrr (6x,4e15.6)
Include these 9 lines (from --- to --- lines) for first grid point only!!
End each grid point entry with a blank line.
End data file with a line of zero's as shown on the next line.
0 0.0 0.0 0.0 0.0 0.0 0.0
-----------------------------------------------------------------------------
V15 sample.LOS.dat Format:
> NEQAIR expects a version number on line 1.
> Descriptive lines follow, with # comments in column 1.
> The last line (no # comment) should contain species names preceded
by titles for data point number n, distance x (cm), and ntot (the
total number density, cm^-3, which is determined and written here
and is NOT expected in the input function file).
Point Redistribution Options:
If the option to reduce the number of points along each line of sight is
employed, this is done based on the magnitude of the distribution of the
gradient of Tv (3rd flow function) with respect to arc length. All of the
function data profiles are interpolated nonlinearly to the redistributed
grid points here, as opposed to requiring the FLOW_INTERP[_2D] step to be
repeated before rerunning NEQAIR_DATA on redistributed line-of-sight data.
The typical function units are so large that these interpolations are
performed in normalized space.
It appears that 3/4ths of the input number of points per line may be the
best compromise, since halving the number tends to lose too much resolution
in the boundary layer of typical forebody lines. For a similar reason, it
appears that the maximum effect of temperature gradient should be invoked
by using a power of 1.0 for the temperature-gradient-based shape function,
where lower power moderates the effect and power = 0. would give uniform
spacing at the other extreme (definitely NOT recommended). Up to 2.0 is
permitted, but don't go that high without scrutinizing the consequences.
Extreme cases (high pressures/thin boundary layers) may not be able to
converge with high exponents anyway. This can mean too much thinning in
the boundary layer, so a hybrid scheme is now an option: the shock region
(defined by the range from the outer boundary to the largest input spacing)
is redistributed by the gradient-base method to the SAME number of shock-
region points, and the rest of each profile employs a 2-sided Vinkour
distribution to obtain the specified total number of points, with the new
wall spacing determined by the ratio of the input to the output number of
points and a heuristic multiplier. This preserves the character of the
boundary layer point distribution, where gradient-based everywhere may not.
Following the hybrid option, a curvature-based option has also been added.
This works with a normalized form of the Tv vs. LOS arc curve. For the
extreme cases of high pressure, this method tends to improve the boundary
layer point distribution, but it does the same sort of thing in the shock
region and therefore tends to put too many points there. Therefore, it
has now been changed to leave a heuristic fraction of the grid points
untouched in the shock region (0.2 currently), giving better resolution
of the boundary-layer-edge region, which appears to be where radiance
losses in NEQAIR occur compared with simple 2x thinning of the CFD points.
That thinning option is now (belatedly) a redistribution option as well.
Most of the above discussion applies to forebody lines of sight (only).
The hybrid and curvature-based methods both assume a shock is present, so do
NOT use either of them on lines in the wake where no shock is encountered.
History:
11/13/13 D.A.Saunders Initial rewrite of a program by Aaron Brandis
that expected LOS data in Tecplot format rather
than PLOT3D multiblock format.
11/14/13 " " Low temperature data outside the shock cause
trouble in NEQAIR. Dinesh Prabhu recommends
suppressing points with Tv < 500 K.
11/15/13 " " Guard against extremely low number densities,
which can affect NEQAIR performance.
11/21/13 " " The arc length outputs now correspond to the
centers of the cells at which the function data
are taken to apply within NEQAIR.
11/23/13 " " An option has been provided to redistribute the
line-of-sight coordinates based on |dTv/ds| and
interpolate all the temperature and species
profiles nonlinearly before writing the lines in
NEQAIR format.
12/03/13 " " GRADDIS[3D]2 now return updated arc lengths
because recovering them from the new x,y[,z]s
is not identical to using the arcs that produced
those new coordinates.
11/24/14 " " Extremely high pressure cases from asteroid entry
studies prompted introduction of a hybrid scheme
for redistributing the data points, as outlined
above (appropriate for forebody data only).
11/25/14 " " Added a same-relative-spacing option that could
be used on forebody lines of sight but is the
likely choice for lines in the wake, where the
hybrid scheme is definitely NOT recommended.
11/26/14 " " x/y/z were not being updated for the Vinokur
part of the hybrid scheme (diagnostic only).
12/01/14 " " Provided a redistribution option that works with
the curvature (in normalized space) of the
(arc length, Tv) distribution.
12/02/14 " " The curvature-based method does what we want in
the boundary layer, but puts too many points in
the shock region. Therefore, simply preserve
the input data points in the shock region and
confine the curvature-based method to the rest
of the profile. Lack of explicit blending of
the two groups of points should not matter much.
12/03/14 " " Raised the heuristic shock-region fraction of the
modified curvature-based method to 0.2 from 0.1.
This puts more redistributed points in the
boundary-layer-edge region, which is where the
main radiance losses appear for NEQAIR compared
with simple 2x thinning of the CFD grid. Talking,
of simple thinning, that is belatedly an option
too now.
12/08/14 " " Averaging of sv(:) as sc(:) allowed an invalid
index of 0 for a wake line of sight.
01/24/14 " " Added the option to output line-of-sight data in
the order required for NEQAIR to integrate away
from the body rather than towards it. This is
required for meteor studies, where the lines are
probably parallel to the X axis, and a "light
curve" is sought (requiring further processing
of the NEQAIR results).
06/04/15 " " The redistribution prompt was missing the 'c'
option; no real change.
03/04/19 " " Started option to output LOS data in the column-
oriented format handled by NEQAIR v15. The legacy
format expected by V14 and its predecessors is
retained as an option, and sample.LOS.dat is used
to determine the kind of output produced.
03/22/19 " " Finished dual output format options and testing.
04/03/19 " " A two-temperature/8 species case encountered an
ambiguity between variable names n and N in the
column-oriented data format of NEQAIR v15. Also,
the index for Tv is not necessarily 3 for this
data format choice.
09/08/19 " " Cosmetic only: N, not Na, in the last comment
above, and one other comment line exceeded 80
characters (printing issue). This utility was
first suspected of letting free-stream temperature
data points through, but the fault proved to be in
FLOW_INTERP's hybrid option.
03/25/20 " " Slight adjustment to skip degenerate lines (nk=2)
in Stardust_Lines output, line_segments.g.
03/30/20 " " Not so slight! Failure to skip degenerate lines
without explicitly reading them was hard to debug.
09/21/20 " " Use of the blackbody BC for Stardust cases (and
meteor cases? TBD) requires lines not starting on
the body to start at low (free-stream) T. Do this
for lines with the first x > ~3*body diameter.
11/29/21 " " Allowed for the Stardust option to perform
gradient-based redistribution if specified. Try
the defaults of 1 and 1 for the exponent and the
shape function only to smooth.
02/27/23 " " The 500 K minimum T for protecting NEQAIR is no
longer needed and isn't good for a blackbody BC.
02/28/23 " " Brett found that the loop looking for Tmin over
the whole line was actually just over i = 1 to 1
after an exchange of %ni and %nk.
Author: David Saunders, ERC, Inc. at NASA Ames Research Center, Moffett Field, CA
Later with AMA, Inc. at NASA ARC.
Download Latest Version
geomlib.tar.gz (58.3 kB)
