Inconsistent color of OBST in smokeview

[obscureed]

I've been seeing some inconsistent coloring of OBST surfaces in smokeview, across several versions (back to FDS6.7.9/SMV6.7.21, and possibly before). This is sometimes just a few OBSTs in large models, where it's difficult to report and not suitable for bug chasing. However, now I have boiled it down to this:

For this model5.fds input file:

&HEAD CHID='model5', TITLE='model5'  /

&TIME T_BEGIN=0., T_END=10. /

&MATL ID= 'CONCRETE',  FYI='Typical Concrete', CONDUCTIVITY =1.4, SPECIFIC_HEAT=0.84, DENSITY=2100    /
&SURF ID= 'WALL', RGB=18,199,18, MATL_ID= 'CONCRETE',  THICKNESS=0.25, DEFAULT=.TRUE.   /

&MESH XB =  -3.000,   3.000,  -3.000,   3.000,   0.000,   3.000, IJK=30,30,15, ID='mesh1' /
&OBST XB =  -1.000,   1.000,  -1.000,   1.000,   0.000,   1.000, SURF_ID='WALL',     /

&TAIL /

How come the central OBST is yellow, not green?

I've labelled this as a smokeview issue, because I think that FDS treats the OBST surfaces as type 'WALL', as specified and as intended.
• The .smv file output from the fds run contains a surface definition 'INERT', which is yellow, then the surface definition 'WALL', which is green (as defined in the RGB=... above), then some other surfaces, some of which are yellow.
• The OBST is listed in the .smv file with s1 etc as surface ID 1 (which I think is the 'WALL' -- we're in a 0-indexed world for smokeview) -- the six 1s at the end of the first line. The surfaces have default color -- the penultimate -1 -- which I think should be the SURF color.

OBST
           1
      -1.00000       1.00000      -1.00000       1.00000       0.00000       1.00000      1   1   1   1   1   1   1                                  
   10   20   10   20    0    5   -1   -1

• I hand-edited the .smv file to recolor all the yellow SURFACE definitions. (I also removed yellow from the MATERIAL definition for CONCRETE, although this is not a documented command, and I don't really expect it to be relevant if I define the color of each &SURF.) As far as I can tell, there are no yellow definitions anywhere in the .smv file. The OBST still shows up as yellow in smokeview -- so where is the yellow arriving from?

As I said, this crops up intermittently and seemingly randomly in large models -- usually, most of the OBST surfaces match the &SURF definition with DEFAULT=.TRUE.. If this is a bug (and I think it is), it makes it unreliable to verify thermal boundary conditions just by inspecting colors in smokeview.

Apologies if this is not a bug and I've missed something. (In the big models, when most of the OBST surfaces match the default SURF definition, it confirms that s1=1, etc, is the second SURF in the .smv file's list -- typically the user-defined default after the yellow 'INERT'. Usually, colorindex=-1, meaning default color, gives the SURF color.)
• I've tried renaming 'WALL' to avoid the possibility of a name overlap.
• Table D.4 of the SMV says that colorindex=-1 gives "default color". If that is intended to result in "default yellow" (RGB= 1.0 0.8 0.4), then I need to find a different test case to show occasions where it inconsistent.

Apologies for the long post. Thanks for your help!

[mcgratta]

This is what I see with the latest nightly build:

[obscureed]

Yes, drat -- I thought it was too easy to find a testcase. The central OBST is yellow in SMV6.7.21-0 and SMV6.10.4-0, at least on my machine.

I'll install the latest nightly build when I get a chance. As a more difficult test, please could you try the attached .smv file (which I've renamed as .txt for attachment)? In those two Smokeviews I just mentioned, it shows as two different kinds of weird. It should be all green. (This .smv file originates from a run in FDS6.7.9, and I have text-edited it to remove lots of DEVCs etc.)

testcase1100.smv.txt

[mcgratta]

[obscureed]

Did you rotate to check? Maybe also toggle pressing 'q'? (The circular fire bed is a different SURF, which is fairly often yellow, although not always. All of the OBST surfaces and MESH wall boundaries should be green.)

Unless there's some definite fix in the latest builds, I think we're just seeing semi-randomness. Not complete randomness -- some OBSTs seem particularly prone to yellowness. That sliver that I think I can see in your image also shows up in SMV6.9.1-0 on a different machine.

[gforney]

smokeview is not identifying that yellow surface as an OBST but as an "other" Surface.

hide Obsts (Show/Hide>Geometry>Obstacles>hidden)

hide "other" Surfaces(Show/Hide>Geometry>Surfaces>other)

we would need to see an input file to see how that surface was defined

make sure you use the nightly build smokeview to generate images

[obscureed]

OK, I've installed the nightly build. It turns out that the .ini file can make a big difference to seeing or not seeing the incorrect yellow. Please try the very simple model5.fds that I started with, but add the attached .ini file (renamed to model5.ini). I can toggle between yellow and green by switching between "perspective" (green) and "orthographic" (yellow).

By the way, there is a mistake in the Smokeview User Guide: it says that the PROJECTION parameter is 1 for perspective, 2 for size preserving. Smokeview behavior is 0 for perspective, 1 for orthographic.

The bug is not only dependent on PROJECTION -- there must be something else going on in the .ini file.

model5.ini_yellow.txt

[gforney]

sorry, I'm not able to reproduce the problem. i installed the latest nightly bundle and downlaoded your .ini file.