Trench Effect

[Kevin]

Has anyone used FDS to model the "trench effect" such as that which
occurred during the Kings Cross fire in London in 1987, or which is
sometimes seen in wildland fires?

[Ken]

Kevin,
My understanding is that the "trench effect" is simply a focusing of
convective and radiative heat energy onto the surfaces. The work I
did on water application rates for road tunnels showed FDS soes a
pretty good job of modeling radiative heat transfer. Without
sprinkler spray, I was getting around 100 kW/m2 which was enough to
ignite the target pile. Wood ignites at 12 kW.m2. If you would like
any further info, please let me know.
Ken

[Kristopher Overholt]

I can't seem to find the input file that I used, but in a fire dynamics class a few years ago I modeled the trench effect in a simplified case following the basic geometry from the King's Cross fire. The model was very simple with stairs, walls to enclose them, open vents on the top and bottom (only passive ventilation), and a HRR vent at the bottom. The trench effect was achieved as described in the King's Cross report, and the flames/flow travelled at an angle that followed the stairwell.

--
You received this message because you are subscribed to the Google Groups "FDS and Smokeview Discussions" group.
To post to this group, send email to fds...@googlegroups.com.
To unsubscribe from this group, send email to fds-smv+u...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/fds-smv?hl=en.

[Kevin]

Thanks for the feedback. I think that there's a dramatic change in the
dynamics of the fire and plume that leads to a more rapid upward
spread than one would first estimate based on classical flame spread
models. It is the unique geometry of the "trench" that brings about
the change in the fire dynamics that then changes the heat flux to the
unburned fuel ahead of the fire. At least, this is my understanding of
it having never tried a calculation like this in 3-D. I did some 2-D
calculations early in my career, but I don't think these calculations
explain the phenomenon completely.

> > sometimes seen in wildland fires?- Hide quoted text -
>
> - Show quoted text -

[Kevin]

Thanks, Kris. Did you repeat the calculation with just a slope and no
"trench"? I saw some YouTube videos from the U of Edinburgh
demonstrating via small scale experiments the change in spread rate as
a funciton of trench angle.

> > fds-smv+u...@googlegroups.com<fds-smv%2Bunsu...@googlegroups.com­>

> > .
> > For more options, visit this group at

> >http://groups.google.com/group/fds-smv?hl=en.- Hide quoted text -

[Kristopher Overholt]

As you said above, I would also expect to see a difference in the results of a classic flame spread model vs. FDS because FDS is resolving for the buoyancy-driven hot flows in this case which are rapidly traveling up and along the trench, while the classic models have built in heat transfer conditions that are not accounting for this.

For my earlier runs, I didn't vary the slope or remove the trench, but I would expect this to affect the speed of the flow due to changes in the buoyancy, the forward heat transfer rate, and thus, the spread rate. It would be interesting to see the results for a variation of the slope and trench parameters.

Kris

To unsubscribe from this group, send email to fds-smv+u...@googlegroups.com.

[Rein]

The small-scale experiments Kevin refers to were conducted by Aled
Beswick from the University of Edinburgh under my supervision as part
of his BEng thesis (aka BSc project in the USA). They are very simple
experiments using balsa wood to illustrate the trench effect and also
flame spread at different slopes. The report and videos can be found
at:

Aled Beswick Thesis:
http://fire-dynamics.info

Videos are also in Youtube:

* The Trench Effect
http://www.youtube.com/watch?v=BJ6VSOkpDYs

* Flame spread at angles of 15, 90, 0 and -15 degrees
http://www.youtube.com/watch?v=V8gcFX9jLGc

G.

On May 26, 2:47 pm, Kristopher Overholt <koverh...@gmail.com> wrote:
> As you said above, I would also expect to see a difference in the results of
> a classic flame spread model vs. FDS because FDS is resolving for the
> buoyancy-driven hot flows in this case which are rapidly traveling up and
> along the trench, while the classic models have built in heat transfer
> conditions that are not accounting for this.
>
> For my earlier runs, I didn't vary the slope or remove the trench, but I
> would expect this to affect the speed of the flow due to changes in the
> buoyancy, the forward heat transfer rate, and thus, the spread rate. It
> would be interesting to see the results for a variation of the slope and
> trench parameters.
>
> Kris
>

> > <fds-smv%2Bunsu...@googlegroups.com<fds-smv%252Buns...@googlegroups.com>

> > ­>
> > > > .
> > > > For more options, visit this group at

> > > >http://groups.google.com/group/fds-smv?hl=en.-Hide quoted text -

>
> > > - Show quoted text -
>
> > --
> > You received this message because you are subscribed to the Google Groups
> > "FDS and Smokeview Discussions" group.
> > To post to this group, send email to fds...@googlegroups.com.
> > To unsubscribe from this group, send email to

[cwood]

Kevin,

I have always thought that the trench effect enhancement was similar
to fire growth in a corner but with the added enhancement of better
air/oxygen entrainment into the increased energy zone for combustion.
In other words, I am suggesting that the free oxygen flowing down the
trench creates enhanced mixing while the angled arrangement provides
for more flame attachment to the angled surface. The "walls" of the
trench enhance the attachment and flame leaning effect. Are you
suggesting some different action from that or some alternative
phenomena? Beswick's work, mentioned by Guillermo, I think
demonstrates this.

[Laurie Odgers]

Kevin,
I'm having some problems getting FDS with openMPI working on our new
machine. I've compiled openMPI and FDS locally using GCC/GFortran and
both work on our two other machines, but this new one produces a
segfault when trying to run the binary through both openmpi and directly.

Here is the output -
[fds-02:08256] *** Process received signal ***
[fds-02:08256] Signal: Segmentation fault (11)
[fds-02:08256] Signal code: Address not mapped (1)
[fds-02:08256] Failing at address: 0x8
[fds-02:08256] [ 0] [0x7d8ed5]
[fds-02:08256] [ 1] [0x7df310]
[fds-02:08256] [ 2] [0x820704]
[fds-02:08256] [ 3] [0x81dd0c]
[fds-02:08256] [ 4] [0x81ec05]
[fds-02:08256] [ 5] [0x82267a]
[fds-02:08256] [ 6] [0x821480]
[fds-02:08256] [ 7] [0x822d3c]
[fds-02:08256] [ 8] [0x824068]
[fds-02:08256] [ 9] [0x821480]
[fds-02:08256] [10] [0x82427f]
[fds-02:08256] [11] [0x81461f]
[fds-02:08256] [12] [0x8146b6]
[fds-02:08256] [13] [0x809b6a]
[fds-02:08256] [14] [0x8099c2]
[fds-02:08256] [15] [0x742c8d]
[fds-02:08256] [16] [0x74320d]
[fds-02:08256] [17] [0x748c54]
[fds-02:08256] [18] [0x74a484]
[fds-02:08256] [19] [0x72c537]
[fds-02:08256] [20] [0x67c314]
[fds-02:08256] [21] [0x688ce7]
[fds-02:08256] [22] [0x667e28]
[fds-02:08256] [23] [0x6548d6]
[fds-02:08256] [24] [0x7a28ee]
[fds-02:08256] [25] [0x7da5e1]
[fds-02:08256] [26] [0x40018a]
[fds-02:08256] *** End of error message ***
Segmentation fault

Is there any way to turn this into a readable stacktrace?

And output of a single cpu from /proc/cpuinfo
processor : 11
vendor_id : GenuineIntel
cpu family : 6
model : 44
model name : Intel(R) Xeon(R) CPU X5650 @ 2.67GHz
stepping : 2
cpu MHz : 1596.000
cache size : 12288 KB
physical id : 1
siblings : 6
core id : 10
cpu cores : 6
apicid : 52
fpu : yes
fpu_exception : yes
cpuid level : 11
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge
mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm syscall
nx pdpe1gb rdtscp lm constant_tsc ida nonstop_tsc arat pni monitor
ds_cpl vmx smx est tm2 ssse3 cx16 xtpr sse4_1 sse4_2 popcnt lahf_lm
bogomips : 5320.57
clflush size : 64
cache_alignment : 64
address sizes : 40 bits physical, 48 bits virtual
power management: [8]

The serial version of FDS works perfectly.

Thanks for your help.

[Laurie Odgers]

I forgot to mention that openmpi is statically compiled into the binary,
and I've tried using minimal optimisation in GCC/GFortran (-O1 instead
of -O3) but it produces the same result.

[Laurie Odgers]

One more thing - the segfault happens even if I dont specify an input file

[Laurie Odgers]

One final update-

I've gotten the following info from strace -

mprotect(0x2ac0fafb0000, 4096, PROT_READ) = 0
mprotect(0x2ac0fad8b000, 16384, PROT_READ) = 0
mprotect(0x2ac0faa3c000, 4096, PROT_READ) = 0
munmap(0x2ac0fa823000, 63860) = 0
mmap(NULL, 1048576, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1,
0) = 0x2ac0fafb2000
open("/etc/passwd", O_RDONLY) = 10
fcntl(10, F_GETFD) = 0
fcntl(10, F_SETFD, FD_CLOEXEC) = 0
fstat(10, {st_mode=S_IFREG|0644, st_size=1624, ...}) = 0
mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0)
= 0x2ac0fb0b2000
read(10, "root:x:0:0:root:/root:/bin/bash\n"..., 4096) = 1624
read(10, "", 4096) = 0
close(10) = 0
munmap(0x2ac0fb0b2000, 4096) = 0
open("/fds/openmpi-1.4.1/lib/libnss_ldap.so.2", O_RDONLY) = -1 ENOENT
(No such file or directory)
open("/etc/ld.so.cache", O_RDONLY) = 10
fstat(10, {st_mode=S_IFREG|0644, st_size=63860, ...}) = 0
mmap(NULL, 63860, PROT_READ, MAP_PRIVATE, 10, 0) = 0x2ac0fb0b2000
close(10) = 0
open("/lib64/libnss_ldap.so.2", O_RDONLY) = 10
read(10,
"\177ELF\2\1\1\0\0\0\0\0\0\0\0\0\3\0>\0\1\0\0\0\0w\4\0\0\0\0\0"..., 832)
= 832
fstat(10, {st_mode=S_IFREG|0755, st_size=3170256, ...}) = 0
mmap(NULL, 5330072, PROT_READ|PROT_EXEC, MAP_PRIVATE|MAP_DENYWRITE, 10,
0) = 0x2ac0fb0c2000
mprotect(0x2ac0fb39e000, 2331800, PROT_NONE) = 0
mmap(0x2ac0fb59d000, 176128, PROT_READ|PROT_WRITE,
MAP_PRIVATE|MAP_FIXED|MAP_DENYWRITE, 10, 0x2db000) = 0x2ac0fb59d000
mmap(0x2ac0fb5c8000, 62616, PROT_READ|PROT_WRITE,
MAP_PRIVATE|MAP_FIXED|MAP_ANONYMOUS, -1, 0) = 0x2ac0fb5c8000
close(10) = 0
--- SIGSEGV (Segmentation fault) @ 0 (0) ---

The only difference between the existing nodes and the current node is
that the old nodes are running centos 4.8 and the new one is running
5.5. Its obviously a problem with libnss_ldap.so.2 but even when I
compile on the new machine it presents the same errors?

Could it be a bug in libnss_ldap.so.2 on centos 5.5?

for the record, LDAP is used for authentication.

[Kevin]

Thanks for the input. Actually, I started this thread because a
colleague asked if there still was interest in the Trench Effect, I
suppose from a modeling perspective. Your explanation sounds
reasonable to me. K