Visibility conditions on FDS+Evac
667 views
Skip to first unread message
enronc
Mar 18, 2011, 8:57:15 AM3/18/11
to FDS and Smokeview Discussions
Dear friends, TimoK,
I am trying to reproduce the evacuation conditions with FDS+Evac of a
tunnel experiment about human behaviour and exit choices . Each
person will choose between an emergency exit or the main entrance in a
smoke filled tunnel (we will put artificial smoke within the tunnel).
I will perform an a priori and a posteriori modeling of the
experiments. I have a set of questions.
Assuming that the smoke distribution is homogeneous during the
experiments, I am simulating the visibility conditions just setting
the parameter &INIT MASS_FRACTION(2) value following one of your
example tests:
&INIT MASS_FRACTION(2)='calculated value' /
&REAC ID = 'CARBONSOOT'
FYI = 'ficticious fuel, 100% soot'
SOOT_YIELD = 1.0
SOOT_H_FRACTION = 0.0
N = 0.0
C = 1.0
H = 0.0
O = 0.0
MASS_EXTINCTION_COEFFICIENT=8700.0 /
In theory, it has to work well for assumed fixed visibility conditions
and without external light sources (otherwise I think it will be not
possible to take into account them within FDS+Evac).
One of the experiments will investigate different conditions of
visibility of the exit signs, so imagining that the environmental
visibility conditions will be fixed, I want to investigate a priori
with FDS+Evac how many people will “see” the emergency exit. I read
on the FDS+Evac manual that “an exit is usable as long as visibility
is larger than half the distance to the exit”. So, I was thinking to
play with the XYZ coordinates, which are used in the exit door
selection algorithm to check if the exit is visible or not.
After the experiments, I will calibrate the position of this point,
but if I want to do an a priori modeling, I would appreciate if you
can suggest me some papers to calibrate in advance the visibility
conditions of the exit sign.
Can you also suggest me an alternative/better way to simulate the
differences in the visibility of the emergency exit apart from my idea
to play with the XYZ coordinates of the Exit? I have read that the
visibility factor C (3 by default for a light reflecting sign) is
another parameter I can play with. How do you think I can use
appropriately this parameter? Can you suggest me some paper on the
topic apart from the Mullholland chapter on the SFPE handbook?
Another part of the experiments will analyse the impact of different
lighting systems. Is it possible to insert external light sources in
some way to modify the visibility conditions within FDS+Evac?
Thank you in advance for your help,
Enrico
I am trying to reproduce the evacuation conditions with FDS+Evac of a
tunnel experiment about human behaviour and exit choices . Each
person will choose between an emergency exit or the main entrance in a
smoke filled tunnel (we will put artificial smoke within the tunnel).
I will perform an a priori and a posteriori modeling of the
experiments. I have a set of questions.
Assuming that the smoke distribution is homogeneous during the
experiments, I am simulating the visibility conditions just setting
the parameter &INIT MASS_FRACTION(2) value following one of your
example tests:
&INIT MASS_FRACTION(2)='calculated value' /
&REAC ID = 'CARBONSOOT'
FYI = 'ficticious fuel, 100% soot'
SOOT_YIELD = 1.0
SOOT_H_FRACTION = 0.0
N = 0.0
C = 1.0
H = 0.0
O = 0.0
MASS_EXTINCTION_COEFFICIENT=8700.0 /
In theory, it has to work well for assumed fixed visibility conditions
and without external light sources (otherwise I think it will be not
possible to take into account them within FDS+Evac).
One of the experiments will investigate different conditions of
visibility of the exit signs, so imagining that the environmental
visibility conditions will be fixed, I want to investigate a priori
with FDS+Evac how many people will “see” the emergency exit. I read
on the FDS+Evac manual that “an exit is usable as long as visibility
is larger than half the distance to the exit”. So, I was thinking to
play with the XYZ coordinates, which are used in the exit door
selection algorithm to check if the exit is visible or not.
After the experiments, I will calibrate the position of this point,
but if I want to do an a priori modeling, I would appreciate if you
can suggest me some papers to calibrate in advance the visibility
conditions of the exit sign.
Can you also suggest me an alternative/better way to simulate the
differences in the visibility of the emergency exit apart from my idea
to play with the XYZ coordinates of the Exit? I have read that the
visibility factor C (3 by default for a light reflecting sign) is
another parameter I can play with. How do you think I can use
appropriately this parameter? Can you suggest me some paper on the
topic apart from the Mullholland chapter on the SFPE handbook?
Another part of the experiments will analyse the impact of different
lighting systems. Is it possible to insert external light sources in
some way to modify the visibility conditions within FDS+Evac?
Thank you in advance for your help,
Enrico
TimoK
Mar 21, 2011, 6:47:16 AM3/21/11
to FDS and Smokeview Discussions
Dear Enrico,
I do not know how your REAC-line will work, but you can
make a small test calculation to see if it works as it
should. Add some output like:
&SLCF PBZ=1.5, QUANTITY='VOLUME FRACTION', SPEC_ID='SOOT' / mol per
mol
&SLCF PBZ=1.5, QUANTITY='DENSITY', SPEC_ID='SOOT' / mg per m3
&SLCF PBZ=1.5, QUANTITY='EXTINCTION COEFFICIENT' / K, m^-1
&SLCF PBZ=1.5, QUANTITY='VISIBILITY' / S = C/K, m
Check that you get the soot density (visibility) that you want.
And the visibility of the exits and doors in FDS+Evac. If the
exits (doors) are regarded as known doors (KNOWN_DOOR_NAMES etc
set on EVAC-lines) then the agents will try to use these doors even
if there is so much smoke that they can not see the exits. They
know that the exits are there. But if there is very much smoke
(visibility
is less than half of the distance) then the agents consider that
they can not use that door because there is too much smoke. They
do not like to inhale so much smoke. Well, this is just a crude
approximation so that the agents try to avoid places where there
is a really bad conditions.
If the exits are not known then the agents should be able to see
the exits or they are not used. Here the soot is used in the
visibility checks. So, you should model the emergency exit as
not known door and put the XYZ so that the agents in the tunnel
can see it. And same for the main entrance if the agents do not
have any bias to choose that door. If the agents like more the
main door than the emergency exit then you could make the main
entrance as known door. The agents will use the known door (main
entrance) if there is also some towards the emergency exit door.
If there is smoke towards the main door but not towards the emergency
exit then the agents choose the emergency exit ("no smoke door").
But if there is some smoke on both doors then the known door
is chosen. "Some smoke" = user input (the default is something, see
the manual) and there is not too much smoke (too much smoke =
visibility
is larger than half the distance to the exit). But note that the
not known doors should be visible through the smoke. The agents
can not use not known doors if they can not see them (well, you can
move the XYZ and play around a little bit, no other exit signage
yet in the model). If an agent has chosen a not known door as its
target exit then this exit has been visible at that point. After that
the exit might not be visible anymore (more smoke or some OBST has
come to block the visibility), but the agent still knows that there
is an exit there. It has seen it previously and the agent has chosen
that exit so it can then be regarded as known exit door and this is
how it is done.
Visibility factor 3.0 is used: Visibility S = 3/K, K is extinction
coeff.
That you can not change. If you want to use some other visibily
factor,
you should scale your soot density.
> Can you also suggest me an alternative/better way to simulate the
> differences in the visibility of the emergency exit apart from my idea
> to play with the XYZ coordinates of the Exit?
Well, if the smoke density in the experiment is constant you can
introduce the smoke like in the "Validation and Verification" page
http://www.vtt.fi/proj/fdsevac/fdsevac_validation.jsp
"Verification of Some Submodels", "Smoke density vs walking speed".
Well, if you are compiling the source code and use the latest
versions (basically FDS6) then that input will not work, but
the corresponding FDS6 style input file is found:
http://virtual.vtt.fi/virtual/proj6/fdsevac/fds6/examples/soot_vs_speed_500.fds
(change the "fds5" to "fds6" in the path name to see the other files
in
FDS6 style, too). Then it is easy to change the visibility by
changing just the smoke density. And there should be a way that
you can initialize some part of the domain with some density and
some other part with some other density, or is there? I have not
been reading the FDS user's guide for a while :-)
There has been some experimental work of the visibility of
different exit signs done in Lund University in Sweden:
http://www.brand.lth.se/publications/#0
Search there for their publications. It might be that some
of the publications are only in Swedish, but that is not
a problem for a Finn, but might be a problem in Italy.
Swedish is the other official language in Finland and
everybody has to learn it in the school.
TimoK
I do not know how your REAC-line will work, but you can
make a small test calculation to see if it works as it
should. Add some output like:
&SLCF PBZ=1.5, QUANTITY='VOLUME FRACTION', SPEC_ID='SOOT' / mol per
mol
&SLCF PBZ=1.5, QUANTITY='DENSITY', SPEC_ID='SOOT' / mg per m3
&SLCF PBZ=1.5, QUANTITY='EXTINCTION COEFFICIENT' / K, m^-1
&SLCF PBZ=1.5, QUANTITY='VISIBILITY' / S = C/K, m
Check that you get the soot density (visibility) that you want.
And the visibility of the exits and doors in FDS+Evac. If the
exits (doors) are regarded as known doors (KNOWN_DOOR_NAMES etc
set on EVAC-lines) then the agents will try to use these doors even
if there is so much smoke that they can not see the exits. They
know that the exits are there. But if there is very much smoke
(visibility
is less than half of the distance) then the agents consider that
they can not use that door because there is too much smoke. They
do not like to inhale so much smoke. Well, this is just a crude
approximation so that the agents try to avoid places where there
is a really bad conditions.
If the exits are not known then the agents should be able to see
the exits or they are not used. Here the soot is used in the
visibility checks. So, you should model the emergency exit as
not known door and put the XYZ so that the agents in the tunnel
can see it. And same for the main entrance if the agents do not
have any bias to choose that door. If the agents like more the
main door than the emergency exit then you could make the main
entrance as known door. The agents will use the known door (main
entrance) if there is also some towards the emergency exit door.
If there is smoke towards the main door but not towards the emergency
exit then the agents choose the emergency exit ("no smoke door").
But if there is some smoke on both doors then the known door
is chosen. "Some smoke" = user input (the default is something, see
the manual) and there is not too much smoke (too much smoke =
visibility
is larger than half the distance to the exit). But note that the
not known doors should be visible through the smoke. The agents
can not use not known doors if they can not see them (well, you can
move the XYZ and play around a little bit, no other exit signage
yet in the model). If an agent has chosen a not known door as its
target exit then this exit has been visible at that point. After that
the exit might not be visible anymore (more smoke or some OBST has
come to block the visibility), but the agent still knows that there
is an exit there. It has seen it previously and the agent has chosen
that exit so it can then be regarded as known exit door and this is
how it is done.
Visibility factor 3.0 is used: Visibility S = 3/K, K is extinction
coeff.
That you can not change. If you want to use some other visibily
factor,
you should scale your soot density.
> Can you also suggest me an alternative/better way to simulate the
> differences in the visibility of the emergency exit apart from my idea
> to play with the XYZ coordinates of the Exit?
introduce the smoke like in the "Validation and Verification" page
http://www.vtt.fi/proj/fdsevac/fdsevac_validation.jsp
"Verification of Some Submodels", "Smoke density vs walking speed".
Well, if you are compiling the source code and use the latest
versions (basically FDS6) then that input will not work, but
the corresponding FDS6 style input file is found:
http://virtual.vtt.fi/virtual/proj6/fdsevac/fds6/examples/soot_vs_speed_500.fds
(change the "fds5" to "fds6" in the path name to see the other files
in
FDS6 style, too). Then it is easy to change the visibility by
changing just the smoke density. And there should be a way that
you can initialize some part of the domain with some density and
some other part with some other density, or is there? I have not
been reading the FDS user's guide for a while :-)
There has been some experimental work of the visibility of
different exit signs done in Lund University in Sweden:
http://www.brand.lth.se/publications/#0
Search there for their publications. It might be that some
of the publications are only in Swedish, but that is not
a problem for a Finn, but might be a problem in Italy.
Swedish is the other official language in Finland and
everybody has to learn it in the school.
TimoK
enronc
Mar 21, 2011, 8:14:30 AM3/21/11
to FDS and Smokeview Discussions
Dear Timo,
Thank you for your comments and suggestions. I have inserted the
Slices you suggested to me to check the visibility conditions and I
found my calculations were ok. I am running this project together with
the team of the Lund University (I am here in Lund now), so I am aware
of their studies (and I know that some of them are in Swedish.. but
they are helping me with the translation).
Due to the conditions of the experiments, I was thinking to set no
known doors (people will be not aware of the exit positions during the
experiments), but I was thinking to set the FLOW_FIELD_ID in the EVAC
namelist with the MESH corresponding to the main entrance (my guess is
that in reality people tend to go straight if they cannot see any door
available). So, I am not using the FED values (we will not intoxicate
people during the experiments!) to decide if an exit is visible or
not, but I just want to reproduce the visibility influence on the exit
choice. I read in the User’s guide that I have to set the parameter
FED_DOOR_CRIT if I want to consider this point as well. My FED will be
always 0 in my simulations, so I read that If FED_DOOR_CRIT> 0.0 then
a door is considered to be smoke free, if the estimated FED index
value for this agent is less than the given value. If < 0.0 then the
absolute value is the
visibility distance (m) which is used by the door selection algorithm
to rank a door as smoke free. I see that the Default is 0.000001. So,
have I to set it as a value lesser than 0 to let use the visibility
distance used by the door selection algorithm?
Of course, after the experiments it will be very easy to calibrate the
exit choice of the occupants by assigning the KNOWN_DOOR parameter.
But the interesting part of my Research is to see the comparison
between my a priori modeling work, the experiments and how then can I
adjust the inputs to let the models work as in reality (a posteriori
analysis).
Just some features that I would love to see in the future versions of
FDS+Evac (Sorry if I am mentioning something that is already available
or you are working in):
- Possibility to implement external sources of light. I was talking
with Prof. Rubini (he is investigating in depth this topic) of the
University of Hull and he explained to me how the influence of
external sources of light could make wrong any calculation of
visibility with CFD models. In the experiments we will try to keep the
visibility conditions as constant as possible and we will try to avoid
external sources of light. But in real tunnels (and building in
general), the conditions are completely different and would be a great
improvement in the model to check the effect of different lighting
systems by using FDS+Evac.
- One part of the tunnel has an incline (a quite relevant slope) that
I have modeled through the &EVSS parameter. I am currently working
with fixed visibility conditions, but I do not know how my
calculations would be reliable if the visibility conditions will be
changing in space (i.e. modeling the changing visibility conditions on
the incline due to the different heights on it). It would be
interesting to check an inclined slice parallel to the incline that
show the changing visibility conditions. It would be also interesting
to check how the different heights of people (due to the incline) will
affect the actual visibility conditions.
About the incline, can you suggest me some reference about the
possible speed reduction on it in relationship to the gradient of the
slope? I did not find so much literature expect the ones that refers
to stair inclines.
Regarding the question to initialize some part of the domain with some
density and some other part with some other density, I read in the
manual that I can set the XB of the initial conditions, but I have not
still tried to set different values of &INIT MASS_FRACTION(2)(i.e. I
set the initial conditions for the whole domain). In theory, in this
way I can simulate the better conditions on visibility due to the
presence of a lighting system on the exit signs or in whatever other
part of the tunnel (I would appreciate if somebody could tell me if it
is possible to do it).
I would appreciate if you can tell me if I am correctly model my
scenarios or if you are thinking there are mistakes/easier way to
represent my scenario conditions.
Thank you one more time (especially for the great job you are doing
with EVAC.. it is incredible how many different scenario conditions
you can simulate with it!)
Cheers,
Enrico
Thank you for your comments and suggestions. I have inserted the
Slices you suggested to me to check the visibility conditions and I
found my calculations were ok. I am running this project together with
the team of the Lund University (I am here in Lund now), so I am aware
of their studies (and I know that some of them are in Swedish.. but
they are helping me with the translation).
Due to the conditions of the experiments, I was thinking to set no
known doors (people will be not aware of the exit positions during the
experiments), but I was thinking to set the FLOW_FIELD_ID in the EVAC
namelist with the MESH corresponding to the main entrance (my guess is
that in reality people tend to go straight if they cannot see any door
available). So, I am not using the FED values (we will not intoxicate
people during the experiments!) to decide if an exit is visible or
not, but I just want to reproduce the visibility influence on the exit
choice. I read in the User’s guide that I have to set the parameter
FED_DOOR_CRIT if I want to consider this point as well. My FED will be
always 0 in my simulations, so I read that If FED_DOOR_CRIT> 0.0 then
a door is considered to be smoke free, if the estimated FED index
value for this agent is less than the given value. If < 0.0 then the
absolute value is the
visibility distance (m) which is used by the door selection algorithm
to rank a door as smoke free. I see that the Default is 0.000001. So,
have I to set it as a value lesser than 0 to let use the visibility
distance used by the door selection algorithm?
Of course, after the experiments it will be very easy to calibrate the
exit choice of the occupants by assigning the KNOWN_DOOR parameter.
But the interesting part of my Research is to see the comparison
between my a priori modeling work, the experiments and how then can I
adjust the inputs to let the models work as in reality (a posteriori
analysis).
Just some features that I would love to see in the future versions of
FDS+Evac (Sorry if I am mentioning something that is already available
or you are working in):
- Possibility to implement external sources of light. I was talking
with Prof. Rubini (he is investigating in depth this topic) of the
University of Hull and he explained to me how the influence of
external sources of light could make wrong any calculation of
visibility with CFD models. In the experiments we will try to keep the
visibility conditions as constant as possible and we will try to avoid
external sources of light. But in real tunnels (and building in
general), the conditions are completely different and would be a great
improvement in the model to check the effect of different lighting
systems by using FDS+Evac.
- One part of the tunnel has an incline (a quite relevant slope) that
I have modeled through the &EVSS parameter. I am currently working
with fixed visibility conditions, but I do not know how my
calculations would be reliable if the visibility conditions will be
changing in space (i.e. modeling the changing visibility conditions on
the incline due to the different heights on it). It would be
interesting to check an inclined slice parallel to the incline that
show the changing visibility conditions. It would be also interesting
to check how the different heights of people (due to the incline) will
affect the actual visibility conditions.
About the incline, can you suggest me some reference about the
possible speed reduction on it in relationship to the gradient of the
slope? I did not find so much literature expect the ones that refers
to stair inclines.
Regarding the question to initialize some part of the domain with some
density and some other part with some other density, I read in the
manual that I can set the XB of the initial conditions, but I have not
still tried to set different values of &INIT MASS_FRACTION(2)(i.e. I
set the initial conditions for the whole domain). In theory, in this
way I can simulate the better conditions on visibility due to the
presence of a lighting system on the exit signs or in whatever other
part of the tunnel (I would appreciate if somebody could tell me if it
is possible to do it).
I would appreciate if you can tell me if I am correctly model my
scenarios or if you are thinking there are mistakes/easier way to
represent my scenario conditions.
Thank you one more time (especially for the great job you are doing
with EVAC.. it is incredible how many different scenario conditions
you can simulate with it!)
Cheers,
Enrico
TimoK
Mar 22, 2011, 4:12:21 AM3/22/11
to FDS and Smokeview Discussions
Hi Enrico,
> experiments), but I was thinking to set the FLOW_FIELD_ID in the EVAC
> namelist with the MESH corresponding to the main entrance (my guess is
Well, that might be a good tactics.
> choice. I read in the User’s guide that I have to set the parameter
> FED_DOOR_CRIT if I want to consider this point as well. My FED will be
> always 0 in my simulations, so I read that If FED_DOOR_CRIT> 0.0 then
> a door is considered to be smoke free, if the estimated FED index
> value for this agent is less than the given value. If < 0.0 then the
> absolute value is the
> visibility distance (m) which is used by the door selection algorithm
> to rank a door as smoke free. I see that the Default is 0.000001. So,
> have I to set it as a value lesser than 0 to let use the visibility
> distance used by the door selection algorithm?
Yes, set it less than zero. The default in the latest source code
in the SVN is using less than zero as the default. (And there are
some other changes also, search for "EVAC_FDS6" in this discussion
forum.)
> Of course, after the experiments it will be very easy to calibrate the
> exit choice of the occupants by assigning the KNOWN_DOOR parameter.
> But the interesting part of my Research is to see the comparison
> between my a priori modeling work, the experiments and how then can I
> adjust the inputs to let the models work as in reality (a posteriori
> analysis).
Well, you can play around with the FED_DOOR_CRIT (less than zero,
i.e.,
visibility) value. If the visibility is better than the given value
(well, the absolute value, because the input is negative...) then
the agent prefers these exits to exits where the visibility is less
than the given value. So, if there is not much smoke between the
agent and the exit ("not much smoke" = visibility larger than the
input value abs(fed_door_crit) ) then this exit is prefered to those
exits where there are some smoke. And if there is really lot of smoke
at some exit then it might be that this exit is excluded and it is not
used at all.
> Just some features that I would love to see in the future versions of
> FDS+Evac (Sorry if I am mentioning something that is already available
> or you are working in):
Well, there are some modifications already there. And some I'm
planning
to do.
> - Possibility to implement external sources of light. I was talking
> with Prof. Rubini (he is investigating in depth this topic) of the
> University of Hull and he explained to me how the influence of
> external sources of light could make wrong any calculation of
> visibility with CFD models. In the experiments we will try to keep the
> visibility conditions as constant as possible and we will try to avoid
> external sources of light. But in real tunnels (and building in
> general), the conditions are completely different and would be a great
> improvement in the model to check the effect of different lighting
> systems by using FDS+Evac.
Well, this is not the first thing that I'm going to do. This sounds
more like things that Glenn do with Smokeview. You should see the
results and see if the light is good for evacuation or not. The agents
will not be using any advanced ray-tracing or similar information.
The only way of doing this would be to calculate the visibility of
different exit signs and other similar stuff (like the doors and
windows etc) and save this information on the (x,y) mesh cells.
The agents could then use this information. But it is not my expertise
to calculate the visibility of different things with different
lighting.
But I could use that information if it would be present at the (x,y)
mesh.
> - One part of the tunnel has an incline (a quite relevant slope) that
> I have modeled through the &EVSS parameter. I am currently working
> with fixed visibility conditions, but I do not know how my
> calculations would be reliable if the visibility conditions will be
> changing in space (i.e. modeling the changing visibility conditions on
> the incline due to the different heights on it). It would be
> interesting to check an inclined slice parallel to the incline that
> show the changing visibility conditions. It would be also interesting
> to check how the different heights of people (due to the incline) will
> affect the actual visibility conditions.
The smoke information at the EVSS incline is taken at the different
z level than the ordinary horizontal evacuation mesh parts. So, the
z_smoke = z_smoke_0 + z_incline, where z_smoke_0 is the level defined
by the HUMAN_SMOKE_HEIGHT (and evacuation mesh XB and
EVACUATION_Z_OFFSET)
parameter. But all agents are using same z and all agents are seeing
through each other. The evacuation geometry is two dimensional. Making
it three dimensional would need much more memory and would slow down
the calcualation also.
> About the incline, can you suggest me some reference about the
> possible speed reduction on it in relationship to the gradient of the
> slope? I did not find so much literature expect the ones that refers
> to stair inclines.
Well, I can not help with this. But there are some marathon races that
are uphill all the way, somewhere in Germany or Austria they run at
the
Alps. You could check the times and how much they go up and see how
much slower they are... And slope speeds depend on on the surface and
also on the shoes, high heels are not very good shoes for large
gradients.
> Regarding the question to initialize some part of the domain with some
> density and some other part with some other density, I read in the
> manual that I can set the XB of the initial conditions, but I have not
> still tried to set different values of &INIT MASS_FRACTION(2)(i.e. I
> set the initial conditions for the whole domain). In theory, in this
> way I can simulate the better conditions on visibility due to the
> presence of a lighting system on the exit signs or in whatever other
> part of the tunnel (I would appreciate if somebody could tell me if it
> is possible to do it).
Well, might be possible to have different conditions at different
zones. Zones are just for the fire part, there are no zones at
the evacuation part. So, you could put some OBSTs with
EVACUATION=.FALSE. (or some HOLEs with EVACUATION=.TRUE.) to make
zones in fire part (your smoke). This way you would keep different
densities at different parts, but the agents in the evacuation
geometry would not see or feel the walls that are needed to construct
the zones.
> I would appreciate if you can tell me if I am correctly model my
> scenarios or if you are thinking there are mistakes/easier way to
> represent my scenario conditions.
Well, as I said, there might be or might not be easier ways to
have your constant smoke densities for the fire part. But for the
evacuation part your approach seems to be fine. Just do the
calculations
with different visibility criterion (fed_door_crit < 0). And if
you can compile the latest SVN release, you could try the
EVAC_FDS6=.TRUE.
style, where the OBSTs for movement are taken at the different z level
that the OBSTs for visibility. But it might be that in your tunnel
geometry this does not matter.
> Thank you one more time (especially for the great job you are doing
> with EVAC.. it is incredible how many different scenario conditions
> you can simulate with it!)
Thanks,
Timo
> experiments), but I was thinking to set the FLOW_FIELD_ID in the EVAC
> namelist with the MESH corresponding to the main entrance (my guess is
> choice. I read in the User’s guide that I have to set the parameter
> FED_DOOR_CRIT if I want to consider this point as well. My FED will be
> always 0 in my simulations, so I read that If FED_DOOR_CRIT> 0.0 then
> a door is considered to be smoke free, if the estimated FED index
> value for this agent is less than the given value. If < 0.0 then the
> absolute value is the
> visibility distance (m) which is used by the door selection algorithm
> to rank a door as smoke free. I see that the Default is 0.000001. So,
> have I to set it as a value lesser than 0 to let use the visibility
> distance used by the door selection algorithm?
in the SVN is using less than zero as the default. (And there are
some other changes also, search for "EVAC_FDS6" in this discussion
forum.)
> Of course, after the experiments it will be very easy to calibrate the
> exit choice of the occupants by assigning the KNOWN_DOOR parameter.
> But the interesting part of my Research is to see the comparison
> between my a priori modeling work, the experiments and how then can I
> adjust the inputs to let the models work as in reality (a posteriori
> analysis).
i.e.,
visibility) value. If the visibility is better than the given value
(well, the absolute value, because the input is negative...) then
the agent prefers these exits to exits where the visibility is less
than the given value. So, if there is not much smoke between the
agent and the exit ("not much smoke" = visibility larger than the
input value abs(fed_door_crit) ) then this exit is prefered to those
exits where there are some smoke. And if there is really lot of smoke
at some exit then it might be that this exit is excluded and it is not
used at all.
> Just some features that I would love to see in the future versions of
> FDS+Evac (Sorry if I am mentioning something that is already available
> or you are working in):
planning
to do.
> - Possibility to implement external sources of light. I was talking
> with Prof. Rubini (he is investigating in depth this topic) of the
> University of Hull and he explained to me how the influence of
> external sources of light could make wrong any calculation of
> visibility with CFD models. In the experiments we will try to keep the
> visibility conditions as constant as possible and we will try to avoid
> external sources of light. But in real tunnels (and building in
> general), the conditions are completely different and would be a great
> improvement in the model to check the effect of different lighting
> systems by using FDS+Evac.
more like things that Glenn do with Smokeview. You should see the
results and see if the light is good for evacuation or not. The agents
will not be using any advanced ray-tracing or similar information.
The only way of doing this would be to calculate the visibility of
different exit signs and other similar stuff (like the doors and
windows etc) and save this information on the (x,y) mesh cells.
The agents could then use this information. But it is not my expertise
to calculate the visibility of different things with different
lighting.
But I could use that information if it would be present at the (x,y)
mesh.
> - One part of the tunnel has an incline (a quite relevant slope) that
> I have modeled through the &EVSS parameter. I am currently working
> with fixed visibility conditions, but I do not know how my
> calculations would be reliable if the visibility conditions will be
> changing in space (i.e. modeling the changing visibility conditions on
> the incline due to the different heights on it). It would be
> interesting to check an inclined slice parallel to the incline that
> show the changing visibility conditions. It would be also interesting
> to check how the different heights of people (due to the incline) will
> affect the actual visibility conditions.
z level than the ordinary horizontal evacuation mesh parts. So, the
z_smoke = z_smoke_0 + z_incline, where z_smoke_0 is the level defined
by the HUMAN_SMOKE_HEIGHT (and evacuation mesh XB and
EVACUATION_Z_OFFSET)
parameter. But all agents are using same z and all agents are seeing
through each other. The evacuation geometry is two dimensional. Making
it three dimensional would need much more memory and would slow down
the calcualation also.
> About the incline, can you suggest me some reference about the
> possible speed reduction on it in relationship to the gradient of the
> slope? I did not find so much literature expect the ones that refers
> to stair inclines.
are uphill all the way, somewhere in Germany or Austria they run at
the
Alps. You could check the times and how much they go up and see how
much slower they are... And slope speeds depend on on the surface and
also on the shoes, high heels are not very good shoes for large
gradients.
> Regarding the question to initialize some part of the domain with some
> density and some other part with some other density, I read in the
> manual that I can set the XB of the initial conditions, but I have not
> still tried to set different values of &INIT MASS_FRACTION(2)(i.e. I
> set the initial conditions for the whole domain). In theory, in this
> way I can simulate the better conditions on visibility due to the
> presence of a lighting system on the exit signs or in whatever other
> part of the tunnel (I would appreciate if somebody could tell me if it
> is possible to do it).
zones. Zones are just for the fire part, there are no zones at
the evacuation part. So, you could put some OBSTs with
EVACUATION=.FALSE. (or some HOLEs with EVACUATION=.TRUE.) to make
zones in fire part (your smoke). This way you would keep different
densities at different parts, but the agents in the evacuation
geometry would not see or feel the walls that are needed to construct
the zones.
> I would appreciate if you can tell me if I am correctly model my
> scenarios or if you are thinking there are mistakes/easier way to
> represent my scenario conditions.
have your constant smoke densities for the fire part. But for the
evacuation part your approach seems to be fine. Just do the
calculations
with different visibility criterion (fed_door_crit < 0). And if
you can compile the latest SVN release, you could try the
EVAC_FDS6=.TRUE.
style, where the OBSTs for movement are taken at the different z level
that the OBSTs for visibility. But it might be that in your tunnel
geometry this does not matter.
> Thank you one more time (especially for the great job you are doing
> with EVAC.. it is incredible how many different scenario conditions
> you can simulate with it!)
Timo
enronc
Mar 22, 2011, 4:26:49 AM3/22/11
to FDS and Smokeview Discussions
Thank you Timo for all your suggestions and advice!
Enrico
Enrico
enronc
Apr 20, 2011, 8:50:36 AM4/20/11
to FDS and Smokeview Discussions
Dear friends, Timo,
I have another question regarding this topic.
You previously said here that: "The agents will not be using any
and I have the values of different VISIBILITY_FACTOR C. Is it
sufficient to insert this different value in a REAC line for changing
the results from the door selection algorithm of FDS+Evac? I mean, is
FDS+Evac taking into account also the VISIBILITY_FACTOR C or just the
smoke conditions?
I have another question regarding this topic.
You previously said here that: "The agents will not be using any
advanced ray-tracing or similar information. The only way of doing
this would be to calculate the visibility of different exit signs and
other similar stuff (like the doors and windows etc) and save this
information on the (x,y) mesh cells. The agents could then use this
information. But it is not my expertise to calculate the visibility of
different things with different lighting. But I could use that
information if it would be present at the (x,y) mesh."
I have done the calculation about the visibility of different objects
this would be to calculate the visibility of different exit signs and
other similar stuff (like the doors and windows etc) and save this
information on the (x,y) mesh cells. The agents could then use this
information. But it is not my expertise to calculate the visibility of
different things with different lighting. But I could use that
information if it would be present at the (x,y) mesh."
and I have the values of different VISIBILITY_FACTOR C. Is it
sufficient to insert this different value in a REAC line for changing
the results from the door selection algorithm of FDS+Evac? I mean, is
FDS+Evac taking into account also the VISIBILITY_FACTOR C or just the
smoke conditions?
TimoK
Apr 21, 2011, 3:51:55 AM4/21/11
to FDS and Smokeview Discussions
Hi Enrico,
For now things are coded as:
! Visibility S = 3/K, K is extinction coeff.
FED_DOOR_CRIT = 3.0_EB/FED_DOOR_CRIT ! Extinction coeff (1/m)
This could be easily changed in new versions, I should just
use the user specified C factor. But the present version is
not doing this. Actually, I just added to my TO DO list that
I'll put to the EXIT and DOOR namelist an optional input
keyword, that is the visibility factor C, so different
doors can have different visibility (reflective exit
sign, a sign with a light bulb, etc).
What is now done vor the visible exits (doors) to check if
they are visible? Calculate (sum) the mass_ext_coff*soot_density
along the bee line towards the door. This does not use any
information on the overall lighting conditions, how lights
reflect from the walls, are the surfaces "black bodies" or
mirrors etc. It just "thinks" that there is one light
source (the exit sign) and there is no other light source
(not even reflective walls) and sees, if the radiation
coming from the exit sign is able to penetrate the smoke
to the position of the agent.
TimoK
For now things are coded as:
! Visibility S = 3/K, K is extinction coeff.
FED_DOOR_CRIT = 3.0_EB/FED_DOOR_CRIT ! Extinction coeff (1/m)
This could be easily changed in new versions, I should just
use the user specified C factor. But the present version is
not doing this. Actually, I just added to my TO DO list that
I'll put to the EXIT and DOOR namelist an optional input
keyword, that is the visibility factor C, so different
doors can have different visibility (reflective exit
sign, a sign with a light bulb, etc).
What is now done vor the visible exits (doors) to check if
they are visible? Calculate (sum) the mass_ext_coff*soot_density
along the bee line towards the door. This does not use any
information on the overall lighting conditions, how lights
reflect from the walls, are the surfaces "black bodies" or
mirrors etc. It just "thinks" that there is one light
source (the exit sign) and there is no other light source
(not even reflective walls) and sees, if the radiation
coming from the exit sign is able to penetrate the smoke
to the position of the agent.
TimoK
enronc
Apr 21, 2011, 6:45:25 AM4/21/11
to FDS and Smokeview Discussions
Thank you Timo,
It is exactly what I have understood. Great to know that you have the
influence of the visibility factor C in your TO DO list.
Cheers,
Enrico
It is exactly what I have understood. Great to know that you have the
influence of the visibility factor C in your TO DO list.
Cheers,
Enrico
Reply all
Reply to author
Forward
0 new messages