FDS Version 4 nearing the end of Testing

[Kevin McGrattan <kevin.mcgrattan@nist.gov>]

All

We're moving towards an official release of FDS/Smokeview
version 4.
Our work in the WTC fires has delayed us more than we
expected. As we
finalize manuals, web pages, etc., could all of those who
have reported
bugs in the new version try the latest. Both Smokeview and
FDS 4 were
posted this week (fire.nist.gov/fds --> Updates). Those who
haven't run
versions 4, could we ask that you take some case you have
created for
version 3 and run it with version 4. The input file
structure has remained
more or less the same with the following exceptions:

1. Restarting jobs is slightly different. In version 4, set
RESTART=.TRUE.
   on the MISC line.

2. Particle and Droplet information is different. See the
new User's Guide
   for details. All the old parameters will still be
accepted, but the
   interpretation may be slightly different.

3. Surface parameters related to the density, specific heat,
etc have been
   expanded and have slightly different interpretation,
especially those
   related to the fuel content of a solid surface.

I have added a useful feature called a HOLE. For example:

&HOLE XB=2,3,5,6,8,9 /

will create a hole for 2<x<3, 5<y<6, 8<z<9. This is useful
for creating
doors and windows. See User's Guide for more info.




The parallel version has been made more efficient as we get
better at
using MPI. Those running in parallel can do the following
test:

1. Put SYNCHRONIZE=.TRUE. on the TIME line.

2. Run a job where the meshes have about the same number of
cells.

3. In the .out file, look at the elapsed wall time between
100 iteration
   printouts. Divide the number by 100, and compare to the
max CPU time
   reported by the individual meshes. On our Linux cluster,
we are seeing
   the CPU/wall ratio between 80 and 95%. Ideally, we want
100%, but there
   are inefficiencies that make this impossible. We'd like
to know how the
   code is running on Windows clusters.



A new researcher has joined our staff. His name is William
(Ruddy) Mell.
He worked here at NIST as a Post-Doc and several years
after, then went to
work at the U of Utah for 5 years and has returned. He has
used FDS for
a variety of special applications, including microgravity
combustion and
forest fires. In the months ahead he will begin fielding
some of your
questions about Direct Numerical Simulation (DNS) and flame
spread.


Assuming we do not run into any serious problems, we hope to
get a 
release version out by mid-May.

Thanks

Kevin



-- 
Kevin McGrattan
National Institute of Standards and Technology
100 Bureau Drive, Stop 8663
Gaithersburg MD 20899-8663

Telephone: (301) 975 2712
FAX:       (301) 975 4052