Body Size and Hole Size (FDS6 Evacuation)

[HF LEUNG]

Hi,

I tried to use FDS6 to simulate an evacuation model with the human body size (FDS's default adult size) larger than the width of the hole. The FDS6 does work but it looks like some humans enter the hole and stick there for a long time before they leave the hole. Did I make any mistake? Please advise. 

The code is attached. Thanks.

Regards,

HF

[TimoK]

There should not be holes smaller than the diameter of the agents in the geometry.
Usually, this means that you should not have holes that are smaller than 0.7 m
in the geometry. If you have these kind of holes, then the geometry is not correct.
The geometry should just contain open floor area as "free space", the other parts
should be OBSTs or should have solid walls so that no agent can go there.

You should block all holes and areas that the agents can not use. In some cases
you have balconies (e.g. in atriums), where agents can move. But in the fire
geometry you do not have there any railings modelled, because the railings
are not blocking smoke movement. You need to say to FDS+Evac, that the
agents should not drop down from the balconies. And you also need to tell
FDS+Evac, where the agents should not go, like small holes in the evacuation
geometry.

So, use:

&OBST XB=...,  EVACUATION=.TRUE., COLOR='RED' /

(Use whatever color you like, but it should be something that you can
see easily. This helps you to see in Smokeview, which "holes" you
have already blocked.

Evacuation geometry is completely different object than the fire geometry.
You just use the fire geometry to make the "initial guess" for the evacuation
geometry. After that, you should construct the evacuation geometry as you like.
The fire geometry => evacuation geometry is just to help the users of FDS+Evac
to construct the evacuation geometry. Usually, most of the fire OBST you want
to be also included in the evacuation geometry. But some fire OBSTs you do not
want. For example, some internal doors that are closed during fire, but used the outgoing
agents. These internal doors should be "holes" in the evacuation geometry. So, you
need to use lines like:

&HOLE XB=...., EVACUATION=.TRUE. /

TimoK

[HF LEUNG]

Hi TimoK,

Thank you for your reply.

What I am trying to model is a train which undershoot/overshoot at the platform during abnormal situation (e.g. fire). That means the train door is not aligned with the platform screen door and this gives a small space for the passengers to escape from inside the train to the platform. This explains why the size of the escape route is smaller than the size of the passengers.

I have revised the FDS input file and re-run the program. It seems the last 17 passengers are not willing to escape from time period of 988s to1200s. Please see the attached result. Not sure why, any advice? Thanks.

Ho Fai

[TimoK]

Well, I'll run the case and see how it goes.

But you can not use FDS+Evac to model special narrow escape routes directly.
You should model a special escape route part differently. You could use a CORR
namelist to construct an effective corridor/stair, that contains one person and
set the lenght and CORR speed factors so that you get a suitable time delay
for your case. The persons are going thorugh the narrow route form the train
to the platform one by one, so you can use: 0.8m width DOOR => CORR with
suitable length and speed factors => DOOR/ENTR 0.8m widht at the platform.
For the CORR you can specify the maximum number of agents inside, in your
case this would be one.

This kind of approach you can use also for ticket counters, etc special escape
routes, where persons go one by one.

TimoK

[TimoK]

You use 0.5m grid in the x and y directions. So, your door/exit widths are
bound due to this dx,dy to: 0.5m, 1.0m, 1.5m, and so on.

Do you want to model both the train and the platform movement?
If so, you should change your EXITs to DOORs like:
(we use now dx=dy=0.5m grid, like yours)

&DOOR ID='TopExit1', IOR=+2, XB=4.5, 5.5, 3.0, 3.0, 0.9, 1.1,
  TO_NODE='TopCorr1', EXIT_SIGN=.TRUE. /

&CORR ID='TopCorr1', MAX_HUMANS_INSIDE=1, TO_NODE='TopExit1PlatDoor',
  XB=4.5, 5.5, 3.0, 3.0, 0.9, 1.1,,FAC_SPEED=0.1, EFF_LENGTH=1.0 /

&ENTR ID='TopExit1PlatDoor', XB=4.5, 5.5, 3.5, 3.5, 0.9, 1.1, IOR=+2, /

And define your mesh so that you have more empty space in the top direction.
Now the mesh ends at y=4m. Set it to 5m, for example. The DOOR 'TopExit1'
makes an automatic OBST behind it, that is one grid cell thick => XB=4.5, 5.5, 3.0, 3.5, 0.9, 1.1,
is now solid obstacle. Change your mesh dx,dy, if you need to model the geometry better.

Now the agents go one by one through the door=>corr=>entr chain. This
takes some time that is:  eff_length / (v0_i*fac_speed)
For an agent with walking speed 1.3 m/s this means: 1.0m / (1.3m/s * 0.1) =7.69s.

Set the eff_length and fac_speed so that you get a value that you want.

It would be best to define the train as its own evacuation mesh and the platform
as a separate evacuation mesh. DOOR=>CORR=>ENTR works then better, i.e.,
less bugs etc. DOORs (with TO_NODE) are basically used to move agents from
one mesh to an another mesh. Use MESH_ID in the keywords, if you have defined
overlapping evacuation meshes, so that FDS+Evac knows to which mesh these
should be applied (XB is not enough to make the mesh unambiguous, where the meshes
overlap). Or you could define the train mesh to be a little bit higher than the platform
mesh (this might be the case or not for you train). This way you do not need
MESH_IDs to be unambiguous.

TimoK

[HF LEUNG]

TimoK,

This task is to estimate the time taken for all passengers to escape from train compartment through the narrow opening between the train door and platform screen door, so it is not necessary to consider the passengers movement in platform area.

It is agreed that the mesh is not correct, I should make it as dx=dy=0.1 to have a better geometry. After this change, none of the passengers can escape. This is probably because the body size (0.5m) is larger than size of the narrow opening (0.4m). I think this is what you said earlier: "you can not use FDS+Evac to model special narrow escape routes directly".

One strange observation. At the beginning, I used DIAMETER_DIST=0, DIA_MEAN = 0.5 for body size and FDS does not show any error message. But when I changed it into DIAMETER_DIST=4, DIA_MEAN=0.489,DIA_PARA=2.614, an error message pops up as "EVAC: Emesh 1 MainEvacGrid has 2 door flow fields".

Why?

Thanks.

Ho Fai

[TimoK]

This is not an error message. It is printed on the stdout/stderr chanel (usually the screen, if
you have not redirected stdout/stderr chanels to some file, e.g.). It just gives some information
to the user. The message means that the evacuation mesh has two doors/exits. I.e., it tells how
many "guiding flow fields for evacuation" are calculated for this mesh. Every door/exit which some
agent could use to go out of this mesh, should have a door flow field. All EXITs have this field,
but the DOORs migh not. The DOOR should have TO_NODE parameter given, if there is a
door flow for this door.

 >  an error message pops up as "EVAC: Emesh 1 MainEvacGrid has 2 door flow

You can not use FDS+Evac to predict, how long does it take to go through a 40cm narrow
opening. For your case, you do not need any evacuation programme to calculate your
train egress. There will be queue at the narrow door. It will be your bottle neck. You should
use some experimental information on how fast the persons go to the narrow opening.

So, make an experiment: Make a mock-up of the train door - platform screen and
have some persons to go through the system. Take a video and see, how much
does it take time. You could do this for the person separately (one person at
a time) and then also in a group mode. I.e., simulate a part of the train car
close to the opening and fill it with persons. This way you can see, what
is the person flow (persons per second).

TimoK

[HF LEUNG]

Thanks TimoK.

I tried to use change from "No Distribution Method" to "Normal Distribution Method" and put the following code, an error message of "FDS was improperly set-up". But when I get back to Normal Distribution, the programme runs perfectly. Can you please advise? Thanks.

DIAMETER_DIST=4, DIA_MEAN=0.5,DIA_PARA=2.61,DIA_LOW=0.44, DIA_HIGH=0.55,

HF

[HF LEUNG]

Forgot to attach the FDS file.

HF

[TimoK]

Your error means that the evacuation initialization part was not succesfull
placing the agents on the given XB area. So, you try to put too many agents
on a too small area. Decrease the number of agents.

FDS+Evac places the initial agents randomly in the XB area and also the
body size and the angle are random. So, if you agent number is close to
the limit, then some runs might be able to place the agents and some not.

You use DIST=4 => normal distribution. It reads just two parameters,
the MEAN and the PARA, MEAN is the mean, and PARA is the sigma
i.e., standard deviation. You can have very large or very tiny persons.
Use DIST=2, which is truncated normal distribution, there the LOW and
HIGH are also used.

TimoK

[HF LEUNG]

It now looks perfectly. Thanks TimoK.

Best regards,

HF