Beam detector over several meshes

[Jonathan Wahlqvist]

Hi,

I'm working with a case with beam detectors over a large area meaning
that several meshes must be used. Problem is that beam detectors
cannot cross mesh boarders. So I came up with a solution that i think
is correct, throwing it in here for anyone to have their say.

I first determined the point where the beam is cut in the mesh
boarder. Then made a new beam stretching from that point to my end
point, a simple example below:

&HEAD CHID='linje3'/
&TIME T_END=120.00/
&DUMP RENDER_FILE='linje3.ge1', DT_DEVC=0.1000/
&MISC RADIATION=.FALSE./

&MESH ID='MESH', IJK=15,30,30, XB=0.00,1.50,0.00,3.00,0.00,3.00/
&MESH ID='MESH02', IJK=15,30,30, XB=1.50,3.00,0.00,3.00,0.00,3.00/

&SURF ID='Fire',
COLOR='RED',
HRRPUA=50.00/

&DEVC ID='BEAM', QUANTITY='path obscuration',
XB=0.00,1.50,1.50,1.50,2.50,2.50/
&DEVC ID='BEAM02', QUANTITY='path obscuration',
XB=1.50,3.00,1.50,1.50,2.50,2.50/
&OBST XB=0.1000,0.70,0.1000,0.70,0.00,0.1000, SURF_ID='INERT'/
Obstruction

&VENT SURF_ID='Fire', XB=0.1000,0.70,0.1000,0.70,0.1000,0.1000/ Vent


&TAIL /

In this case the first beam will go from x=0 to 1.5 and then x=1.5 to
3.0.

Then I started the calulations and after looked in the devc.csv.

To add up the two beams resulting in the same as if it was one beam in
one mesh i calculated the following:

B= -ln((1-"obscuration beam 1")/1)+(-ln((1-"obscuration beam 2")/1))

total obscuration = 1-e^(-B)

What I do is calculate the obscuration to optical density and then add
them together and transform it back to obscuration.

Did a couple of test runs and calculations comparing to a single beam
and mesh case (the one beam covering the same path as the two
together) and it seems to give approximately the same answer, small
differences do occur but i think this is due to small differences in
time steps.

Any comments or any other ideas to solve this problem? The bad thing
about this method is that when the beam detector is to activate a fire
vent for example, you have to do the simulation one time to get the
time to detection, and then another one with the fire vent activating
at this given time.

/Jonathan

[Kevin]

This seems like a good strategy to me. You might even try using the
"beam_detector" verification case, breaking it up into two meshes.

The reason why we do not allow beam detectors to cross mesh boundaries
is simply to limit the complexity of parallel calculations. It is very
difficult to pass information efficiently between computers running
MPI. Especially true when the information is not limited to the mesh
boundary. Let us know if this works for you and then we can include
this as guidance in the Users Guide.

Thanks

[Stephen Olenick]

Might be worth posting something to the issue tracker asking for an
enhancement on controls. As far as I know, you can't do a 'SUM'
control or equation-based unit conversions in FDS. If you could, you
could put in an input to convert the obscuration to optical density
for each beam in FDS as you have done by post-processing and have a
control that sums the two together and opens a fire vent at a specific
value. I don't know how hard that is for the developers to add in,
but it would allow you to open vents with a complete beam across
multiple meshes without having to run it twice.

On Jun 2, 9:57 am, Jonathan Wahlqvist <r3jonwa...@hotmail.com> wrote:

[Jonathan Wahlqvist]

Ok, tried the beam_detector verivication test and it turned out to
work perfect, no difference at all in the devc-output file.
Did a minor update to my previous post though, had to many minuses :)
Should be:

B= ln((1-"obscuration beam 1")/1)+ln((1-"obscuration beam 2")/1))

total obscuration = 1-e^(B)

instead.

Posted the *.fds-files and devc-output files here:

http://groups.google.com/group/fds-smv/web/verification_case_beam_detector.7z