Closed gforney closed 9 years ago
gforney
commented
9 years ago Timo will take a look at this. However, before he does, you must simplify the case.
Trying to debug this file is like looking for a needle in a haystack.Original issue reported on code.google.com by mcgratta on 2011-05-19 12:07:47
gforney
commented
9 years ago Well, I'll look this, but could you specify that in which
main evacuation mesh these things happen? I.e., I would
like just to model that mesh, easier to see things in SMV
and faster for CPU.
TimoK
Original issue reported on code.google.com by tkorhon1 on 2011-05-19 13:46:30
gforney
commented
9 years ago Well, one quick suggestion:
You have too long names for the MESH ID:s, try to use
no more than 20 characters (well, might be that 26 or 30
is ok but to be safe, use less than 20 characters).
And do not use space characters, use underscore. This
might not help but it is safer so.
The first 30 characters are used to distinguish the
different meshes. Well, it seems that this is not the
problem, but use still shorter names for the meshes.
And there seems to be some VENTs missing:
&MESH ID='Main Evac Scala B piano terra', IJK=40,50,1, XB=53.20,62.70,24.15,35.55,0.50,0.70,
EVACUATION=.TRUE., EVAC_HUMANS=.TRUE., EVAC_Z_OFFSET=0.50/
&MESH ID='To 0floor exit B', IJK=40,50,1, XB=53.20,62.70,24.15,35.55,0.50,0.70,
EVACUATION=.TRUE., EVAC_Z_OFFSET=0.50/
Do you have here two doors that you want the agents to use
using the door selection algorithm? Then you should have one
main evacuation mesh that has a VENT at both doors, and two
additional evacuation meshes (EVAC_HUMANS=.FALSE., the default)
that have just a VENT at one of the doors.
Now you have just the "outflow" VENTs without any MESH_ID, so the
vents will end up in all meshes. For those vents that are in a
main evacuation mesh that have also additional "door flow"
evacuation meshes, you should use MESH_ID. E.g., two VENTs
at the different doors with MESH_ID='Main_E_mesh', and one
vent with MESH_ID='Emesh_Left_Door', and one vent with
MESH_ID='Emesh_Right_Door'.
Check the "floors" of your building are "sealed" and that
they have just one sealed part that have the outflow vents.
"floor" = main evacuation mesh and this should have walls
everywhere (mesh boundaries are ok if they are at the outer
walls so that no "air" can flow inside the building from
outside).
If you floor contains two or more zones that are sealed form
each other, then you should define different main evacuation
meshes for these parts. Well, if you have a door connecting
these parts in the real geometry, then you could put
HOLE with EVACUATION=.TRUE. there and the two parts
belong now to the same "sealed" space.
TimoK
Original issue reported on code.google.com by tkorhon1 on 2011-05-19 14:12:11
gforney
commented
9 years ago Hi,
Check the attached figure: x=19.26-19.86 for
the "hole", so it is just 60 cm wide. Some agents
can be that wide or close to it. They do not fit
through that "hole". Make that part wider. Note that
the agents need a little bit wider "hole" than their
body width, because there is a social force acting
between the walls and the agent. 80 cm wide path
ways should be fine ususally, less than about 60 cm
are problem. Check the other floors also for similar
things. Use "g" to activate the grid lines in SMV
and Show/hide=>Labels=>Grid locations. And by pressing
"z" you can get rid of the grid lines on the xy-plane.
See the smokeview quide for more, you might need to press
"m" to move between the meshes.
Use HOLE with EVACUATION=.TRUE. to make more space for the
agents. This will not change the fire calculation geometry.
Well, I check the mesh 9 (z=11.25,11.45) and here you can
see the flow vectors (load=> vector slices) that the vectors
are going inside the walls at some points, see the attached
figure TK4_Mesh9Problem.png. You can see that you have a vent
(and a door) that is in a sealed box. (Note, that I do not
have any EVSS, it is better to see everything at the main
evacuation mesh z positions for debugging.)
Well, it seems that you should take away the the door=>door
connection to the stairs, just put there a hole. You have
similar staircase in this floor at a different place, where
you have just a hole between the floor and the stair landing.
My test case was different, I just took the vent, which is
in the staircase, away and then the flow field of that
floor was better, no vectors going inside the walls.
Well, "no vectors" means than in the areas where the agents
are. There are some sealed spaces in that floor, but these
do not have agents nor vents, so they are usually no problem.
The new flow field is attached and it seems to bee quite fine.
So, the flow field should be fine, if you take the door=>door
(and the corresponding vent) away and make there a hole and
have the vent just at the stairs.
Well, these things are going to be a little bit better when
FDS 6 is out there. Easier user input and the stairs should
be quite easy to model with STRS.
Add also SLCF output for all (main) evacuation meshes. And
see them in Smokeview that the vectors are not going inside
the walls (where they should not). If they do, the problem
is usually the "sealed spaces" problem.
I'll change the status of this issue to "AltInputs". Note,
that then this issue is "closed", so it is not shown in
the "open issues" search. If you still have problems with
your input, search for "all issues" so you will find this
issue and we can continue with it.
TimoK
Original issue reported on code.google.com by tkorhon1 on 2011-05-19 15:41:57
_
_
_gforney
commented
9 years ago Ciao Timo, thanx a lot for your time!
I've read a lot from u on this website..and i'm quite your fan!:D
About the hole at the bottom mash i noticed that but hadn't time to fix it because
the stair problems kept hardly my attention.
For the stairs, i tried to remove the vent inside the sealed staircase having as result
a similar vector display as u had, but probably i didn't understand well what u mean
because once removed outflow vent the agents doesn't go trought the stairs and remain
all around the first door! I'll post a pic of that.
Then i hadn't a "door to door" connection but a "door to enter" one and if i remove
the "to node" connection FDS give me the "door to node error".
Maybe i didn't understand it well because of my not so good english knowledge, so if
u can attach the FDS file u've run i can read it line to line and discover where i'm
doing mistakes.
Sorry for lettin'u loose your time, ciao, Teo.Original issue reported on code.google.com by teodorico.centobuchi@hotmail.it on 2011-05-19 23:10:34
_
_gforney
commented
9 years ago P.s. how u managed to have the white mesh backround color instead of default yellow?
it seems more clear to see objects and vectors than default one.
Thanx again.Original issue reported on code.google.com by teodorico.centobuchi@hotmail.it on 2011-05-19 23:47:56
gforney
commented
9 years ago for comment 6:
Menu: Show/ Hide -> Geometry -> Surfaces -> [Hide] ExteriorOriginal issue reported on code.google.com by gregor.jaeger on 2011-05-20 05:35:11
gforney
commented
9 years ago Well, let's make this Issue once again "Open" so
now it has "MoreInfo" status. The program probably
works as it should, just the user input is a little
bit problem. This is also an issue for the user
manual, it should be better. But I'm not going to
update the FDS 5 based user manual anymore. In FDS 6
there will not be any "outflow" vents etc anymore, just
EXIT and DOOR lines. But the problem of "sealed" evacuation
domains inside one main evacuation mesh will be a problem still.
I do not want to add "ZONEs" like in the fire calculation,
because this will mesh up the fire and evacuation meshes
things. It is easier to say in the manual that one main
evacuation mesh should correspond to one "evacuation zone"
which has all parts connected. One should have many main
evacuation meshes per one floor of the building, if there
are areas where you can not go to other areas without
crossing some wall (a door=>door connection is also
considered to be a wall, well, there is always an
OBST behind the "outflow" vent, so there is a wall...)
And the +z and -z surfaces of the evacuation meshes are
showing up by the default yellow colour, so do as Gregor
told, hide the exterior surfaces (the new versions of
Smokeview). In the old versions of Smokeview (say, 9 moths
old or older) these were called as "exterior vents" or
something similar and they are find in show/hide=>vents
or something like that. A technical note why these surfaces
are the default yellow. The evacuation meshes has k=1, i.e.,
one grid cell in the z direction. And there are mirror
boundary conditions at the -z and +z surfaces, so SMV shows
these boundary conditions. So, the FDS flow solver is "tricked"
to calculate a two-dimensional solution for the flow, where the
equations and other stuff is actually for a three-dimensional
flow solver. I just added this show/hide exterior surfaces
to my To Do list, so I should remember to write it to the
manual (or I could also try to "hack" the dump.f90 source
code so that in the .smv file something else is written for
the evacuation meshes...).
> Then i hadn't a "door to door" connection but a "door to enter" one
Well, the flow vectors going inside walls and, thus,
agents banging their heads to a wall does not depend
on the DOOR nor ENTR lines. DOOR=>DOOR and DOOR=>ENTR
are the same for the FDS flow solver. It does not see
any DOOR or ENTR, it just sees the walls (OBSTs + HOLEs)
at the DOOR and ENTR positions and the "OUTFLOW" VENTs.
Well, I'll check the 9th mesh (the one at z=11.25-11.75)
so that it is working correctly and add my comments in the
input file. You can then do the same for the other meshes,
if they have similar things. Note, that I'm not doing
anything for the other meshes. Your picture "4_4315.png":
The door=>entr where the agents are should be removed and
there should just be a HOLE, no OBSTs nor VENTs. And you
put a VENT at DOOR=>entr in the stairs. For my test I took
that vent away, because it was easier to do so for a debugging.
Then I got nice vectors, so there probably is no other "bad"
vent in that mesh.
NOTE: mesh 'Main Evac Terzo piano', IJK=300,226,1, XB=19.51,79.31,12.73,46.98,11.25,11.45,
has:
Delta X
(79.31-19.51)/300
.19933333333333333333
Delta Y
(46.98-12.73)/226
.15154867256637168141
It is better to use XB and IJK so that these are some "nice"
numbers like 0.15, 0.20, 0.25, 0.30, and so on. Then you know
better your door widths etc, because then you know how FDS will
change everything to the underlaying grid. So, in your case the
door widths are not those that you give in your input file. FDS
moves the XB to the grid and the door width will change somewhat.
If you can not find "nice" IJK and XB, you can extend the XB a little
bit outside of your outer walls (if some outer wall is the mesh
boundary then you should put an OBST there, of course).
TimoK
PS. Do not matter on your English, I'm not a native speaker. And
it is good to use your English as much as possible, it will
get better all the time.
Original issue reported on code.google.com by tkorhon1 on 2011-05-20 08:00:46
gforney
commented
9 years ago Hi,
search for "Timo:" in the input file. I made some
changes and added some comments. Now the z=11.25-11.75
should work, but the z=7.8,8.0 has plenty of problems.
Other z levels I did not check.
Ciao,
TimoK
Original issue reported on code.google.com by tkorhon1 on 2011-05-20 10:20:09
gforney
commented
9 years ago Hi,
Timo, thanx for your explanation, i understood what u mean, i tried to remove the door
at staircase and run the simulation (i've put thoose doors to avoid agents toexit from
the staircase and use the stairs in the internal open space). I've also run the .FDS
file u modified.
The vector i have on new simulation are kinda same as the ones you had (no vector going
inside walls but just vectors directing on the wall).
It goes better but still 2 agents crossed wall at 11.25-11.45 mesh.
I let the EVSS in place because for my researches i need the time they could need to
go out trought the stairs.
Now, running your model with no EVSS all works fine, running similar model with EVSS
there are problems. So i'm starting to think that the flow coming inside the staircase
is disturbed by the flow of agents coming down from the stairs turning the corner.
Can be possible?
I've attached vectors pic, TK4_0033 is from your model, 5DT_4353 from mine.
Removing Doors has also caused a collateral effect: now there is a corner with vectors
heading inside the corner and agents stay trapped (the other 2 pics).
Ciao, Teo.Original issue reported on code.google.com by teodorico.centobuchi@hotmail.it on 2011-05-21 05:03:44
_
_
_
_gforney
commented
9 years ago Another little question, i've put in also the fire meshes (4 equal meshes IJK=100,171,104)
and tried to run an openmp simulation with 4 threads on same PC (quad core).
Now, after 12 hours it shows that are calculated just 12.50 seconds, the simulation
is 500seconds, it means that it would need something like 500 hours? It seems to me
a little bit much. Am i doing some mistakes?
Thanx again, Teo.Original issue reported on code.google.com by teodorico.centobuchi@hotmail.it on 2011-05-23 05:27:12
gforney
commented
9 years ago Well, you have 1,778,400 cells in one fire mesh. This is
much and it needs much CPU time.
The other things I'll look later today. One thing
that you could try is to make the evacuation mesh
dx and dy something like 25 cm, 40 cm, or 50 cm,
depending on your main corridor and door widths.
Or you could try to model your staircases using the
STRS method, see the guide and the examples at the
web page
http://www.vtt.fi/proj/fdsevac/fdsevac_validation.jsp
and there
"Validation by comparison against experimental results"
"Stairs Office Building: fire drill Type 1, Type 2, Type 3"
And the "Type 3" is using the STRS thing.
Or you could use the EVSS etc, but make a small main
evacuation mesh for each staircase at the same z level
than the main floor. Use then MESH_ID at your doors and
entrs and "outflow" vents in these small stair meshes,
so that these things do not go to the large floor mesh.
The different main evacuation meshes can overlap, but
then you should use MESH_ID so that the input is
unambigous.
Also in your case, you could devide the floor two
different main evacuation meshes, if you want to
keep the -y part and +y part away and connect them
using door=>door/entr things.
TimoK
Original issue reported on code.google.com by tkorhon1 on 2011-05-23 07:06:26
gforney
commented
9 years ago Check for "TimoNew:" in the attached input file.
I just removed the unnecessary internal door=>entr
connections at the z=7.8-8.0 mesh. The problem for
you might be that the left staircase is having the
exit quite close so the agents go to the exit and
not use the stairs going down. If you want to change
how the agents use the stairs and the exit then you
need to define two additional door flow evacuation meshes
for this main evacuation mesh (EVAC_HUMANS=.FALSE. which
is the default). Then you can give KNOWN_DOOR_NAMES on the
evac-lines so that the agents either use the exit or the
stair door going down. See the manual for details how
to define the additional door flow meshes and how to
define the "outflow" vents for these meshes. Note, that
you should use MESH_ID on both the main evacuation mesh
vents and on the door flow meshes.
Now the main evacuation mesh flow field is looking nice.
If you put any of the "internal" doors back to the geometry,
then you should devide your main evacuation mesh to smaller
meshes, which can be overlapping. But each mesh should contain
just one "sealed" part of the floor. All vents (and doors) should
be in this one "sealed" part only. The mesh will have other parts
of the floor in the geometry also, because your "sealed" parts
are not rectangular, but this does not matter as long that you
have the "outflow" vents just in one sealed part.
If you want that the agents use the door selection algorithm
then you should add additional door flow evacuation meshes
for all the doors and exits at this floor.
TimoK
Original issue reported on code.google.com by tkorhon1 on 2011-05-23 09:21:14
gforney
commented
9 years ago Well, I'll close this issue. It was fixed by changing
the input file. The errors in the original input file
were the common ones:
1) evacuation geometry should not have smaller holes
than about 65 cm so that the largest agents can
also go through these.
2) the (main) evacuation mesh should be one sealed zone
only. If there are more sealed zones in one floor of
the building then more main evacuation meshes should
be used fo this floor.
3) Using the STRS model the internal (and external) staircases
are easier to model. A STRS has its own main evacuation mesh
and there is no problem with the sealed stairs and other parts
of the floor.
TimoK
Original issue reported on code.google.com by tkorhon1 on 2011-06-08 09:08:27
gforney
commented
9 years ago I'm sorry, i didn't write for long time, but i had some life's issues to solve before
going on with my project.
Now it seems i've solved every problem :
1)removed all non necessary doors
2)created specific mash for every staircase (nomore agents pass trought walls)
3)vents of specific meshes are linked to that mesh so doesen't disturb the main floor
vectors
4)tried to use bigger mesh cells but wasn't usefull(bigger cell means thiker obstruction
beetween door-vent and enter and obstruction must be all inside cell so with 10cm standard
walls means that some doors were causing some problems to agent flows) so i came back
to initial mesh cells size
Now my last question is : (on second floor there is an exit "exit5" and on the relative
staircase there is a door who lead to that exit "door 2P-exit".now that exit in reality
isn't a real exit but direct just on the roof of lower structure, is just a secure
place were to wait for help and in my plans able to contain max 25-30 people.
The problem is that more than hundred agents directs there! I've tried to use the "known
exits method" but seems it really doesen't work since all agents i try to direct to
"exit 4" are going trought staircase A to "exit5" or "exit 2".
The question are : there is a way to influence the path choose of agents? and there
is a way to limit the agent flow trought a door?
Thanx.
I attach the fds file .Original issue reported on code.google.com by teodorico.centobuchi@hotmail.it on 2011-08-08 17:04:27
gforney
commented
9 years ago Hi,
Well, I will run your case, but first some fast answers:
Q1: there is a way to influence the path choose of agents?
A1: Well, you can not easily to modify the path of the
agents towards some specific door. But you can change
the target door of the agents quite easily. You can
force some agents to go towards some specific door
always, even if there are other doors (or exits)
at this floor level ("floor level" = main evacuation mesh).
Q2: and there is a way to limit the agent flow trought a door?
A2: Well, there is no way of saying that the door should be
closed when some number of agents have gone through it.
But you could make there a small space behind the door (or
where ever in the geometry where you have some space to
put some artificial room) so that there can not be too
many agents in that small room.
The other way is to define: DOOR => CORR => dummy ENTR
where the dummy entry is placed so that there is not
enough space to put an agent there. The CORR has
parameter MAX_HUMANS_INSIDE, so this should do
the job. Note that there is (probably) input
check that you can not place the ENTR inside a
solid. So, you should have one "empty" grid cell
in front of the ENTR, it is enough to have the
grid cell in front of the XB empty, i.e., the
input check is done for the mid point of the XB.
TimoK
Original issue reported on code.google.com by tkorhon1 on 2011-08-09 09:37:19
gforney
commented
9 years ago Hi,
see the attached file, search for "Timo:".
Now there might still be too many agents
going to the exit5, because the agent
density is quite high in the stairs and
some agents are pushed to the 'door 2P-Exit'
and go through it. If you do not want this
to happen, then you should change the geometry
so that the agents are not pushed to that door.
Make that door more to the left.
Or you could combine the staircase mesh and exit5 mesh,
so you then have just a hole, where you now have the
'door 2P-Exit' => 'ENTR 2P-Exit'. This would be the
best way to model your case. If you do this, then you
should remove the 'door 2P-Exit' from the known door list,
of course. And in this case you should place the XYZ of
the exit5 so that the agents inside the staircase can not
see it. If they see it, they might aim towards exit5,
not towards 'door 7.8 A'. Well, they will see the door 7.8 A
at some point and it is closer than exit5, so they should
choose 'door 7.8 A', if exit5 is not known. But to be sure,
put the XYZ so that it is not seen from the staircase.
TimoK
Original issue reported on code.google.com by tkorhon1 on 2011-08-09 12:57:44
gforney
commented
9 years ago Hi,
I tried to follow suggestions written in the .fds file but as u said now all agents
are going out using exit5. Encountered couple of problems: 1)fds won't start till i
didn't remove the "TAU_FALL_DOWN" command at PERS line. (maybe i used an uncorrect
syntax or not supported by my fds version)
2) i use pyrosim as graphic interface and it doesen't allow me to add doors to the
known doors list, it means i should type it manually for every EVAC line i got?
Then i removed the "scala2P-left and down" meshes and extended the "scala 2P" mesh
to the left till the Exit5 , gave at exit5 an XYZ placed out of the building and now
no agent choose the exit5...so seems that works!!
My last question is, due to the fakt that the A staircase has very high agent density,
how i can manage to direct some of them to the other staircase? just using the known
exit or maybe play with visible points XYZ??
Thanx very much.
Teo.
P.s. i attach the last fds fileOriginal issue reported on code.google.com by teodorico.centobuchi@hotmail.it on 2011-08-10 04:46:57
gforney
commented
9 years ago Hi,
1)
The tau_fall_down was my mistake. I did use some more recent
version of the source code. The 5.5.3 on the download page
does not have this problem.
2)
Well, Pyrosim writes the ".fds" file to the disk, so
open the file in a plain text editor, like notepad or
wordpad (or emacs or vi if you have unix/linux style
things installed in your windows).
" XYZ placed out of the building ": Well, that is not a
good tactics. It should be in front of the door, you could
put it a little bit to the +y direction from the middle of
the door. Actually, it should not matter too much, if the
agents see this, because when they see this they also see
the door=>entr at the stair flight and the stair flight
door is closer than exit5, so they should choose that
one, if exit5 is not the known door. Well, some
queueing time is added to the door choise also, but
the exit5 is so much further than the door at the
stair flight, so queuing does not matter.
"Then i removed the "scala2P-left and down" meshes and extended the "scala 2P"
Well, you should add two additional door flow meshes for your
extended exit5+staircase main evacuation mesh. Now you have
to different target doors in the exit5+staircase mesh: the
exit5 and the door at the stair flight. Easiest for you would
name the meshes as "scala2P-left and down" like before, but
for the scalea2P-left OUTFLOW VENT should be moved to the
exit5. The outflow vent at the stairflight (scala2P-down)
is already there at the correct place. Then you should
modify the EXIT line of exit5. There should read
VENT_FFIELD='scala2P-left', because the additional mesh
that has the vent at exit5 is having this name.
After these modifications the agents from the EVAC
lines that have exit5 as known door (known_door_probs
is one, not zero) will go to exit5, not down the
stairs. The other agents will go down the stairs.
How to guide agents to different staircase (staircase B?).
Well, you should give the known_door_names (and probs)
for the doors leading to the other staircase (the door=>entr
doors) on those EVAC lines, which you want. Well, you could
also give the door=>entr door of the staircase A on the
known door lists.
Well, perhaps in your case you should give both the doors
to staircase A and staircase B as known_door_names (and probs)
on the EVAC lines of the corresponding floor. This way the
agent will go (mainly) to the closest staircase.
For example,
&DOOR ID='door 7.8 B', XB=57.60,57.60,32.65,34.15,7.80,8.00, XYZ=57.50,33.40,8.00,
MESH_ID='Main Evac scala b 2P', IOR=1, TO_NODE='ENTR 7.8 B', EXIT_SIGN=.TRUE.,
KEEP_XY=.TRUE., RGB=255,0,51/
This is leading from a floor to stairs, so give this as known
door for the agents that are generated on this floor (i.e.,
EVAC is at this floor).
Well, "go to closest exit" above means:
Go first to the closest known visible staircase door.
If no such door, go to the closest known door (not visible now).
If no known doors at this main evacuation mesh, then go to
closest visible door. If no visible doors, then there are
no doors available, so can not choose any door. Then the
default behaviour is to follow the "flow field" of the
main evacuation mesh which will lead to some door.
So, the agent tries to choose the "best" door that
is inside the current main evacuation mesh. This
means that the exit1,...,exit5 given as known doors
do not matter until the agent is in the main evacuation
mesh, where the doors are. You should give the all
internal doors leading to, say, exit1 as known doors,
then the agents can follow the route to exit1.
TimoK
Original issue reported on code.google.com by tkorhon1 on 2011-08-10 08:02:25
gforney
commented
9 years ago Good morning,
following your last suggestion i finally let it run properly!! :)
Added the "scala 2P left and down" meshes and assigned to 2 groups of agent colour
red and yellow and the known exit5 input, then seen that the red and yellow agents
going out by the exit5!! :) Still having some questions :
1) That should work for every exit i will add at some floor? i mean if i add an additional
exit for example at 3rd floor main mesh"main evac terzo piano", then i've got to duplicate
the mash like "main evac terzo piano-left", create 2 vents for the exit then add the
vent field at the exit door and all agents having the known new exit prob=1 will go
out by that exit?
2)I gave the known door probs =1,00 for the red ones and =100,00 for the yellow ones
and seen that both direct in the same exit so which is the right imput? 1 or 100? are
thoose just boolean values like 1=yes and 0=no or can be used also fractions like 0,5?
3)I know i can use the text editor to change the .fds imputs, i do it often, but having
a list of about 50 or more different agents groups it is kinda slowly to edit every
single EVAC line.Then knowing i've got to run like 3 simulations(i mean with different
amount of agents) for every fire location (about 3 or 4), rewriting everytime all 50
lines will be kinda hard. :( Btw i've started to do that but will take few time.
4)I use the openmp_win64 fds version since i use a single quad core machine and doesen't
support the TAU-FALL-DOWN command, but i think that there isn't a newest openmp version.
Thaks a lot,
Teo.
i attach the last file.Original issue reported on code.google.com by teodorico.centobuchi@hotmail.it on 2011-08-11 03:48:20
gforney
commented
9 years ago Answers:
1) Yes, if you have more than one exit/door in a main
evacuation mesh, then you need to make additional
"door flow" evacuation meshes for each of the doors.
And, of course, you should have outflow vent at each
of these doors in the main evacuation mesh (use MESH_ID).
But you should also have one vent in every "door flow"
evacuation mesh, at the corresponding door.
Note: In FDS 6, these things are automatic. You just
define main evacuation mesh and the DOOR/EXIT lines.
Then the program creates automatically the outflow
vents and the additional meshes that it needs.
And if you want that the agents regard the two different
exit as much, then you should give both of them as
known doors (usually with probs=1.0). The known_door_probs
is basically a shorthand for many EVAC lines. Say,
known_door_probs=0.5 means that in the initialization
phase, when the agents are generated, it is randomly
decided if an agent knows this door or not. And the
random probability is the know_door_probs. So, if you
want that 50 % of the agents know some exit, then
give probs=0.5. You would get about (its random)
50 % of the agents knowing this door. But you could
also give two different EVAC lines with half of the
number of agents, one having probs=1.0 and the other
one probs=0.0 (or the door is not at all in the
known door list).
2) It is a probability and mathematicians define the
probability between zero and one. It is not per cent.
So, any real number in the interval 0.0 - 1.0. If
you give larger than one then it is same as one.
3) Well, you could do EVAC lists in Excel (or other spreadsheet).
Then you just copy-and-paste them to the text file. Then
you can use the Excel formulas, macros or whatever to speed
up your editing work. And replace string is also nice, but
this should work in a text editor also.
4) Well, I think that the openmp_win64 is FDS 5.5.3. The tau_fall_down
is much later thing. I usually update my source code every day
and compile. Someone who compiled the source code was having some
issue with the tau_fall_down some times ago. But, as I recall it
now, it was much later than the release of FDS 5.5.3, so it is
not FDS 5.5.3 thing. The FDS 5.5.3. is the latest official realeased
version of FDS. This means that the verification and validation (V&V)
is done for that version. The later versions (which you must compile
yourself) are not gone through the official V&V process (yet).
Let me know, if you have still some problems. See also
the thread on the discussion forum
"some problems for evac":
https://groups.google.com/group/fds-smv/browse_thread/thread/3f744a5c470c7233/ab1c744a7a9e8447#ab1c744a7a9e8447
These are the first steps that one should check,
if the input file for evacuation is not working
as you would like.
TimoK
Original issue reported on code.google.com by tkorhon1 on 2011-08-11 07:48:03
gforney
commented
9 years ago Good morning,
I've updated all the EVAC lines with know door probs giving to the agents of each floor
the prob=1 for the door on the staircase B. But as usual 80% of the agents of all floors
head to the staircase A, now the A staircase is usually used by the building workers
only , the B stairs are main public enter of the building, so for being near the reality
the B stairs should be full and the A stairs not.(at least according to the fact that
people tend to exit from building using the path they walk for go inside the building).
Should i hide the XYZ of the B enter-staircase doors?
Sorry for disturb,
Teo.Original issue reported on code.google.com by teodorico.centobuchi@hotmail.it on 2011-08-12 06:13:44
gforney
commented
9 years ago Well, I'll run your latest file and see, how you
will get more agents to staircase B. I'll just check
the floor, where exit5 is. Making it to work there
should be an example for the other floors.
TimoK
Original issue reported on code.google.com by tkorhon1 on 2011-08-12 07:37:23
gforney
commented
9 years ago Hi,
Check the attached input file, search for "Timo:".
I attached also an ".ini" file for Smokeview.
Below are the main points:
Timo: You have two staircases at the floor mesh, the doors to these stairs are:
Timo: 'door 2P A' and 'door 2P B'. You need additional door flow evacuation meshes
Timo: for the these doors:
Timo: Below are your outflow vents, I transfered them to here. But now these
Timo: vents need mesh_id, bacause you want both vents only at the main evacuation
mesh:
Timo: Then you need the vents for the additional door flow meshes, one vent per each:
Timo: You should also add VENT_FFIELD='DoorMesh_2PA_2nd_piano'.
Timo: This tells that which additional door flow field is leading to this door.
Timo: You should also add VENT_FFIELD='DoorMesh_2PB_2nd_piano'
So, if you have more than one door at a floor (here floor means
a main evacuation mesh) then you should have an additional door
flow evacuation mesh for each door. Then you should use
MESH_ID everywhere, because you have now overlapping meshes.
The main evacuation mesh should have OUTFLOW vents at every
door. The additional door flow meshes have just one OUTFLOW
vent at the corresponding door. And the DOOR/EXIT lines should
the have VENT_FFIELD that tells the name (ID) of the correct
additional door flow evacuation mesh.
TimoK
Original issue reported on code.google.com by tkorhon1 on 2011-08-12 13:14:56
gforney
commented
9 years ago Good morning,
now seems that all works fine except for couple of agents who run in circle at "secondo
piano mesh" probably because having scala B door as known exit and they find the door
blocked by other agents who wait to enter, then try to find someother door but again
directed by flow fild in the blocked door.
But after 3/4 circle the door is empty so they going down.
If i will have someother problems may i ask again?
Thanx for all Timo, i owe u a favor!! :)
Have a good summer, Teo.Original issue reported on code.google.com by teodorico.centobuchi@hotmail.it on 2011-08-15 05:24:17
gforney
commented
9 years ago Nice that you got it more or less working.
Well, the summer is more or less over here in
Finland, we are more or less shifting to autumn
already.
TimoK
Original issue reported on code.google.com by tkorhon1 on 2011-08-18 12:50:27
Original issue reported on code.google.com by
teodorico.centobuchi@hotmail.iton 2011-05-19 03:13:47