EVSS with stairs in y direction

[wnst]

Hello, everyone

I have some problems on the setting of Evss connecting the ground floor and the first floor. I wonder if you could help me out?

Referring to the MonStairs example, I changed the direction of EVSS from IOR =+1 to IOR=+2, which connects the Ground floor to the first floor.

Although the evacuation is calculated successfully, the movement of egress at the inlet of the EVSS on the Ground floor is unexpected.

The egress goes up to the first floor as soon as it moves into the Door on the Ground floor. It seems that the egress has not walked on the stairs.

Part of the fds file is shown below, and the whole is attached. 

----------------------------------------------------------------

&DOOR ID='GF_2_Stairs', IOR=+2,

      FYI= 'Comment line',

      VENT_FFIELD='ToStairsUp',

      KEEP_XY=.TRUE.,

      COLOR='PINK', EXIT_SIGN=.TRUE.,

      TO_NODE= 'Stairs_2_Up',

      XYZ=44.5, 9.0, 1.0,

      XB= 42.0,47.0, 10.0,10.0, 0.9,1.1, /

&ENTR ID='Stairs_2_Up', IOR=+2,

      FYI= 'Comment line',

      COLOR='CYAN',

      XB= 42.0,47.0, 20.0,20.0, 4.9,5.1,, / 

---------------------------------------------------------------- 

&EVSS XB=42.0,47.0, 10.0,20.0, 4.9,5.1, IOR=-2, ID='WideStairs',

      FAC_V0_UP=0.4, FAC_V0_DOWN=0.7, FAC_V0_HORI=1.0,

      HEIGHT=0.0, HEIGHT0=-4.0, MESH_ID='FirstFloor' /

 Railings

&OBST XB=41.8,42.0,  10.0,20.0, 4.9,5.1, /

&OBST XB=47.0,47.2, 10.0,20.0, 4.9,5.1, /

----------------------------------------------------------------  

Thanks in advance.

Wnst 

from Sendai, Japan

[TimoK]

Well, do the basic "debugging" measures:

Set EVACUATION_DRILL=.TRUE. and use the Show/Hide menu of
Smokeview to see things nicely.

Set the SLCF output files for the evacuation meshes.

Comment the EVSS out, i.e., take  the "&" away. Or just change
the EVSS HEIGHT0 and HEIGHT equal to zero, so you will see
everything at the "correct" main  evacuation  mesh z-level. See, if
things are going like they should. e.g., the agents jump from one
z-level (mesh) to an another z-level (mesh) there, where you want
them to change "floors" (two main evacuation meshes). Everything
is done at the two-dimensional main evacuation meshes so you
should be able to project the true 3-dimensional geometry to
2-dimensional planes. EVSS are just changing the z-coordinates
to show things better in Smokeview and, of course, changing the
speeds of the agents (and doing some other 2-dim <==> 3-dim
slope transformations that are needed in the equations). But the
geometry is still the 2-dimensional evacuation mesh planes.

Below some copy-and-paste from an another discussion thread here  at
the forum:

The evacuation geometry should  be "sealed" so that you have walls surrounding
your region, where you have the agents. The evacuation mesh boundary is
considered to be a wall. If you get warnings that some agents are removed
from the calculation then the usual case is that you have placed your
OUTFLOW vents and door/exit lines not correctly. See, if you have
the same XB for the door/exit and OUTFLOW vent lines.

Read also:
http://virtual.vtt.fi/virtual/proj6/fdsevac/documents/Readme_InputErrors.txt
(FDS+Evac homepage => "Documentation" subpage => How to debug
your input file: Readme_InputErrors.txt)

Hints:

1) MISC-namelist: "EVACUATION_DRILL=.TRUE." so you can
   have the fire meshes there. They are just not read in.

So, do just the evacuation geometry so you see it better.
And use Smokeview menus to show the evacuation geometry
better, see below:

5) In Smokeview: Show/Hide menu => Geometry => Surfaces
   => take the "Exterior" away. Then you do not see the
   boundary planes of the evacuation meshes and this
   helps you see the evacuation geometry much better.

   Play around also with the other show/hide geometry
   options, I like "show obsts as solid+outline".

TimoK

[wnst]

Hi, Timo

Thank you for your advice.

I managed to change the direction of EVSS.

As I want to do the simulation for a 3-storey building where the agents on the 2nd floor go to the 1st floor and the  agents on the ground floor also go to the 1st floor.

These two Evsses have the same projected plan. The movement of the agents are strange. If these Evsses are located in different places, the simulation has quit well result.

Could you help me out?

By the way I also saw the velocity slice in smokeview, the velocity is about 0~5 m/s 10^-7.  The unit is  "10^-7 " , which means the velocity is very, very small. Why it is so small?

Thank you!

Wnst

2012年5月23日水曜日 16時39分58秒 UTC+9 TimoK:

[TimoK]

First, the easy answer:

SLCF and the evacuation mesh flow fields: The FDS flow solver is used
to "paint arrows to the floor that are leading to the (exit)doors". This
uses two-dimensional (horizontal) meshes and a fictitious fluid that is
sucked out of the building at the door locations. So, one gets directional
arrows towards the exit doors and the agents use these arrows to
navigate towards the doors. Real fluid flow could be turbulent, e.g.,
having vortices, and these are not wanted in the arrows that lead
to the doors. Thus, the velocities are kept very small and the viscosity
and other fluid parameters are changed also so that there will not
be any voritices in the fictitious flow. The magnitude of the velocity
vectors is not used in FDS+Evac, just the direction of the vectors
is used. But I still want to see the magnitude of the velocity in Smokeview,
because then the vectors close to the OUTFLOW vents are usually having
"large" magnitude and are shown as read. Easier to see these positions
in Smokeview. The other possibility would have been to normalize the
vetros to a unit lenght, but this is not so nice for "debugging of the
evacuation flow fields".

The second problem:

   GF => 1st floor  <= 2nd floor

Let's assume that you have intermediate landins, so one between
GF and 1stF, and the other between 1stF and 2ndF. And then
you have four stair flights, of course: like GF => flight1 => landingA =>
flight2 => 1stF and 1stF <= flight3 <= landingB <= flight4 <= 2ndF

If you are in an aeroplane, you will see that the landings A and B have same
XB (x,y components) and also flight1 and flight3 and similarly the flights 2 and 4.
So, you should decide to which "floors" (main evacuation meshes) you will put
these things that have same XB (x,y components, well, "same" means also that
some parts of the objects have same x,y). There can only be one
[http://en.wikipedia.org/wiki/Highlander_%28film%29]. So, you should put some
things to GF, some to 1stF and some to2ndF. There are many ways you can do
that, below is one:

  GF: flight1, landingA, DOOR_A => ENTR_A(at the 1st floor mesh)

  1stF: ENTR_A, flight2 (and now your are already at the 1st floor mesh, so
          you need no door=>entr at the top of the flight2)

  2ndF: flight4, landingB, DOOR_B => ENTR_B(at the 1st floor mesh)

  1stF: ENTR_B, flight3 (and now your are already at the 1st floor mesh, so
          you need no door=>entr at the bottom of the flight3)

I like this way, because now you have the DOOR=>ENTR at the middle
of the stairs, i.e., far away from the point where the two human flows
will merge at the stairs at the 1st floor level. The DOOR=>ENTR should
be kept away from these kind of merging areas, if possible. Why?
Because of the simple ENTR (same for DOOR if it is used to "come out
direction") algorithm. ENTR sees, if there is empty space enough in front
of it and if there is then it places an agent there. Otherwise, the agent
just waits at the ENTR and is at the main evacuation mesh, where the
"sending DOOR" is, so the ENTR side, i.e, the next mesh, does not
know anything about this agent => no forces or other information is
passed through entr=>door connections.

Hope that this will  help you.

TimoK

[wnst]

Timo

 

Thanks for your information.

I haven't succeeded in the simulation of my stair case because of the confusing to the setting of DOOR and ENTR. 

There are some other questions about the evacuation between rooms and stairs.

1.One Evacuation mesh for one floor?

As it is suggested that one main evacuation floor for one floor of the building,and the XB of  door flow evacuation mesh should be the same with the main evacuation floor.

How to set the different door flow evacuation for different doors in different rooms in one floor? 

2. Can the DOOR and ENTR be applied between Rooms?

As to the example (DoorFlowExample), there is only a hole between the room and corridor. Is it possible to place a real door there and set  the open time for the door?

3. The color of agent

As the color of agent's cloth can be changed, if the paints color can also be changed?

Thank you.

Wnst

2012年5月22日火曜日 18時49分32秒 UTC+9 wnst:

[TimoK]

1: See the example in the manual "10. Sample Input Files". There are
two doors per one floor. See how they are done.

2: If you want to model the "internal doors" like a office room door to a
corridor, this is not easy. With FDS5.5.3 evacuation module you might
have some kind of tricks to do this so that you do not need to define
main evacuation meshes for each room and each corridor separately.

3. Just the color of the body ellipsoid could be changed. If you want to
change some other parts then you should modifu the "objects.svo" file
at the FDS7FDS5/bin directory (where fds5.exe etc are).

TimoK

[wnst]

Hi, Timo

 

Many thanks for your reply.

 

Yes, I want to model the "internal doors" for  a office room door to a corridor.

As is shown in the attached figure, I was succeeded in using a Hole between one room and the other.

If it is possible to replace the hole by DOOR, ENTR or others?

 

Could you tell me the details about the tricks?

 

Thank you in advance.

 

 Best Regards

 

Wnst

 

 

2012年6月4日月曜日 18時19分26秒 UTC+9 TimoK:

[TimoK]

If you want to put a DOOR => ENTR where you have now just a hole, this
can be done. There are many ways, but best would be:

  - You need two main evacuation meshes for your floor. One for the
    "left room" and one for the "right room". So, copy-and-paste your
     main evacuation mesh line and rename the meshes "nicely",
     e.g. 'RightEmesh' and 'LeftEmesh'. Change the IJK and XB so
     that you do not have too much right hand side room in the left
     hand side mesh and similarly for the right hand side mesh. The
     main evacuation meshes can overlap, but it is better to define the
     meshes as touching, if possible. This way one does not have
     so many error opprotunities...

  - You should use MESH_ID='xxx' keywords at the DOOR, ENTR, EVAC, etc
    lines so that FDS+Evac knows to which main evacuation mesh these
    things should go, if the XB fits both main evacuation meshes (i.e., XB is
    at a place, where the two main evacuation meshes overlap).

 - Place on OBST where you have now a hole, if needed. If your main evacuation
   meshes have outer boundaries at this point, then this is not necessary.

 - Put the DOOR to the right mesh and the ENTR at the right mesh and add a VENT
   to the right mesh (at the DOOR XB). Well, if you have many doors at the right mesh,
   then you need the VENTs at the main evacuation mesh and the corresponding door
   flow evacuation meshes.

If you do not want two different main evacuation meshes, you could fool the programme
a little bit. Have the hole at the main evacuation mesh. But define an additional door flow
evacuation mesh for your new DOOR. And put on OBST with EVACUATION=.TRUE. and
MESH_ID='your new door flow mesh name' and put here also outflow VENT with
the same mesh ID. And put also the ENTR at the correct position. Now in the main
evacuation mesh, there are no OBST at the door position, so that the agents do not
see a solid wall at the door position, i.e., they can go through the door. The main
evacuation mesh flow  will be as previously. But your new additional  door flow evacuation
mesh has just one outflow vent at the door position, so the flow lines will go towards this
door at the right room. The OBST that is needed for the outflow VENT is at this additional
evacuation mesh and the agents can not see or feel it, so they can use the door=>entr
connection as they should.

But the first type is better. It is better to have two separate main evacuation meshes.
The CPU usage will be less due to the fact that the agents at different meshes do
not interact (a double loop is shorter).

TimoK

[wnst]

Hi, Timo

Many many thanks.

I managed to test the  DOOR => ENTR case according to your detail instruction.

What I want to do is about the evacuation in a multi-story building.

I will try it later, while referring to the stair example from EVAC homepage.

Many new problems will be faced.

Hoping your continuous help.

Thank you  in advance.

Wnst

2012年6月11日月曜日 21時29分25秒 UTC+9 TimoK:

[wnst]

Hi, Timo

I managed to simulate the Door   => ENTR in one floor case last week.

But the agents do not move in the new model with 2 floors.

I am confused about how to divide the main evacuation mesh, especially at the following places. Could you give any advice?

1. As to divide the main mesh for one floor, is it means I have to make a one-floor meshes (main evacuation mesh), then divide main evacuation mesh into the right room mesh , the left room mesh, and the stair mesh.

2. The flow evacuation mesh for every door is needed, and VENTs for doors also need the main evacuation mesh. In my case, what is the main evacuation mesh for VENTs, the one-floor mesh, or the divided main evacuation mesh?

3. When I make a vent for a door, "Vent is not behind an OBST " is often pointed out, because a HOLE is defined for the door. If there is no HOLE, there is no error message.

 How to define the HOLE, VENT, and DOOR?  Should I define the VENTs, DOORs, OBSTs and HOLEs in the same evacuation mesh? 

Thanks

Wnst

2012年6月11日月曜日 21時29分25秒 UTC+9 TimoK:

If you want to put a DOOR => ENTR where you have now just a hole, this

2012年6月11日月曜日 21時29分25秒 UTC+9 TimoK:

2012年6月11日月曜日 21時29分25秒 UTC+9 TimoK:

2012年6月11日月曜日 21時29分25秒 UTC+9 TimoK:

[TimoK]

1. As to divide the main mesh for one floor, is it means I have to make a one-floor meshes (main evacuation mesh), then divide main evacuation mesh into the right room mesh , the left room mesh, and the stair mesh.

So, you have two main evacuation meshes in one floor (right and left ones).
 

2. The flow evacuation mesh for every door is needed, and VENTs for doors also need the main evacuation mesh. In my case, what is the main evacuation mesh for VENTs, the one-floor mesh, or the divided main evacuation mesh?

The door flow evacuation meshes have the same XB and IJK than the main evacuation meshes. So,
you door flow evacuation meshes should also be divided to the right and left ones. Left ones for those
doors that are on the left side and so on...

 

3. When I make a vent for a door, "Vent is not behind an OBST " is often pointed out, because a HOLE is defined for the door. If there is no HOLE, there is no error message.

FDS5 wants that a VENT is on a solid OBST. If you have DOOR => DOOR/ENTR connection, the incoming
door should have a VENT. So, you need an OBST behind this incoming door so that you can place a VENT
there. And you do not need a hole here if you are using a door => door/entr connection. These types of
connections move the agents through the walls (usually through the different floors).

 

 How to define the HOLE, VENT, and DOOR?  Should I define the VENTs, DOORs, OBSTs and HOLEs in the same evacuation mesh? 

You should define the hole,vent,door, etc on the left evacuation mesh(es) for the doors that are on the
left side. And similarly for the right side.

TimoK

[wnst]

Hi, Timo

Thanks for your quick reply.

According to your advice, I made a complicated case. 

There are 4 rooms in one floor, the left one, the right one, the middle one and the stair room, with the interior walls between rooms. I intended to make the agents in the left room move  to the middle one, then to the stair room. So do the agents in the right room. 

But the agents only move in the left and right room, they can not go from one room to the other?

It seems I made some mistakes.Could you tell me where is (are) the problem(s) ?

1.These 4 rooms are in four different evacuation meshes, as well as the flow meshes for different doors in different room. And there is no evacuation mesh for the whole floor. 

2.There is no hole for each the Door=>ENTR., The DOOR and ENTR are with the same geometry and at the sample place, i.e, they have the same XB.  I tried to make a hole for the door in the left room, but  "Vent must be attached to a solid obstruction " is pointed out.

3.There are interior walls behind the ENTR. Maybe they block the agents' movement. 

4.There are some warning at some ENTRs , which shows "mid point facing solid".

5. How to make the agents go from one main evacuation mesh to the other.

Last week, I  succeeded in such a case  without setting the door flow mesh.

Could you give some help?

Thank you.

Wnst

 

2012年6月18日月曜日 17時43分01秒 UTC+9 TimoK:

[TimoK]

Well, let's see. Usually you do not put door => entr/door connections within
a floor, but you can do this. The door => entr/door is not as physical than
just a hole, the human flows are not as good throught a door => entr/door
connection than in a simple hole. But dividing the floor to many main evacuation
meshes and having the door => entr/door connections give more options for the
door choice algorithms that is implemented in FDS+Evac (and much more work
to write the input file...). The usual advice is to put door => entr/door connections
there, where they interrupt the human flow least. It might be that this is not always
an easy task and some compromises should be made.

But now back to your case: Four meshes in one floor, one mesh = one room case:

Left room = RoomL, one door leading out: Just the main evacuation mesh needed, MeshL

Right room = RoomR, one door leading out: Just the main evacuation mesh needed, MeshR

Middle room = RoomM, one door leading out: Just the main evacuation mesh neede, MeshM

Stair room = Stairs, MeshS a STRS mesh (or no mesh at all, if CORR is used). If EVSSs are
used, then the upper part of the stairs could be in the RooM mesh (change the XB and IJK accordingly)
and the lower parts would be in the lower level mesh. But let's assume that a STRS mesh is used.

Agent paths: RoomL => RoomM => Stairs
                    RoomR => RoomM => Stairs

Left means -x direction, Right means  +x direction and we suppose that the
doors between the right and middle room and left and middle room are
moving in the x-direction, so that the doors and vents have x1=x2 and the
door width is in the y-direction (i.e., width = y2 - y1). Hopefully you understand
what I'm trying to tell :-)

Door and outflow-VENT definitions:

 RoomL door out: ID='DoorL',  MESH_ID='MeshL', XB=xxx, etc. TO_NODE='EntrL2M', IOR=+1,
 RoomL, vent: MESH_ID='MeshL', XB=xxx, EVACUATION=.TRUE. SURF_ID='OUTFLOW',...
 Possibly an EVACUATION=.TRUE., MESH_ID='MeshL', OBST if the vent needs it.

 RoomR door out, ID='DoorR',  MESH_ID='MeshR', XB=xxx, etc. TO_NODE='EntrR2M', IOR=-1,
 RoomR, vent: MESH_ID='MeshR', XB=xxx, EVACUATION=.TRUE. SURF_ID='OUTFLOW',...
 Possibly an EVACUATION=.TRUE., MESH_ID='MeshR', OBST if the vent needs it.

 RoomM door out, ID='DoorM',  MESH_ID='MeshM', XB=xxx, etc. TO_NODE='MeshS', IOR=+2 (towards +y direction in this example)
 RoomM, vent: MESH_ID='MeshM', XB=xxx, EVACUATION=.TRUE. SURF_ID='OUTFLOW',...
  (now a STRS mesh is assumed for the stairs, see the examples on the web page).

 RoomM door/entr in left: ID='EntrL2M', XB=xxx, MESH_ID='MeshM', IOR=+1,...
 RoomM door/entr in right: ID='EntrR2M', XB=xxx, MESH_ID='MeshM', IOR=-1,...
And ENTR (or "incoming doors") do not need any outflow vents.
Note: If you would use DOORs instead of ENTRs, then you should change the IORs.
The IOR is "pointing" out of an ENTR, but it is "pointing" towards a door.
 Stairs, The STRS mesh finds the incoming doors automatically, if they are touching the stairs
 mesh boundaries, so no need to make explicitely a door in the MeshS mesh. If you are using
 the simple CORR, then you do not have any stairs mesh at all, just the 'DoorM' would have
TO_NODE='YourCorrName'

So, put the OBST with EVACUATION=.TRUE. (and correct MESH_ID) where you put
your OUTFLOW VENTs. Add also PERMIT_HOLE=.FALSE. to these lines and
THICKEN=.TRUE. so that you do not accidentally take these OBSTs away and
that they are at least one grid cell thick. Add also COLOR='RED' (or some other
color name) to show them clearly in Smokeview.

Hope, that this helps.

TimoK

[wnst]

Hi, Timo

 

Thank you very much.

I managed to do the case with different rooms in one floor.

I made some mistakes in ENTR's setting.

First, the ENTR should be placed in the nextroom mesh., which means the XB of ENTR can be different with that of DOOR.

Second,  the FLOW_FIELD_ID was also considered. If there is only one door in the mesh, FLOW_FIELD_ID can be neglected.

Thus, the agents moves from the left room to the Exit in the stair room.   Thanks!

 

As to the OBST for OUTFLOW VENTs, I have a question.

There are 2 VENTs for one door, one is for the main evacuation and the other is for door flow grid

For example, RoomL door  ID='DoorL', DoorL has one Vent with MESH_ID='MeshL', and the other with MESH_ID='toDoorL', which shows the door flow grid.

But I only made one OBST, is it OK?  Which vent needs this OBST?

 

Thanks in advance.

 

Wnst

 

2012年6月19日火曜日 17時31分22秒 UTC+9 TimoK:

[TimoK]

First, the ENTR should be placed in the nextroom mesh., which means the XB of ENTR can be different with that of DOOR.

Second,  the FLOW_FIELD_ID was also considered. If there is only one door in the mesh, FLOW_FIELD_ID can be neglected.

If you have touching meshes then the XB could be same for both door => entr
if the door => entr is at the mesh boundary. If this is case then the XB will fit
both meshes and you should specify MESH_ID on the DOOR and ENTR lines.
Well, you could also have this case, if your meshes are overlapping as they might
be, if you geometry is not easily constructed using rectangles. For overlapping
(and touching) mesh cases you need to specify MESH_ID, because just the XB
is not enough to specify  unambigously the mesh.

And, if you have a main evacuation mesh, where you have just one outgoing
door (or exit) then you need just one flow field so no need to any other
additional door flow evacuation meshes. And then you need not to specify
the FLOW_FIELD_ID on the DOOR (or EXIT) lines, because the main
evacuation mesh flow field is used by default. And in this case the main
evacuation mesh flow field is guiding all agents towards this door, because
there is just one "outflow vent" in the main evacuation mesh. If you would have
more doors/exits, then the main evacuation mesh would have "outflow vents"
at all door/exit positions and the agents following the main evacuation mesh
flow field would not go to a specific door/exit.

All of this will be much easier when the FDS6 is released (or you compile
the latest source code...). There will just be the main evacuation mesh and
the flow fields are just calculated for each door/exit separately, so there is
no "flow field that will lead to all doors". But the general idea is the same,
the "outflow vent" is placed at a door and then a fictitious 2-dimensional
flow solution (without any vortices) is found so that you will get "painted arrows
on the floor that will guide the agents towards a specific door".

As to the OBST for OUTFLOW VENTs, I have a question.

There are 2 VENTs for one door, one is for the main evacuation and the other is for door flow grid

For example, RoomL door  ID='DoorL', DoorL has one Vent with MESH_ID='MeshL', and the other with MESH_ID='toDoorL', which shows the door flow grid.

But I only made one OBST, is it OK?  Which vent needs this OBST?

Yes, you need two vents. It is the way how it was programmed. It would have
needed some more programming work to make things more automated, so
the copy-and-paste thing was left to the user. The main evacuation mesh should
have "outflow vent" at all doors and exits where the agents might go. This will
produce "painted arrows that lead to some door according the flow solution".
Even if the agents are not ever using the main evacuation mesh flow field
(they are using the door flow fields, when they have a specific target door
dictated by the door selection  algorithm), you need the "outflow vents" at
the main evacuation mesh. The "outflow boundary condition" lets the agents
go "inside a solid", i.e., they do not see an OBST at the door. This is done
differently in the latest source code (will be in FDS 6), but in FDS 5.5.3 the
evacuation part still needs these vents.

One OBST is enough, if you do not need to specify MESH_ID. So:

 &OBST XB=xxx, EVACUATION=.TRUE., /

is making an OBST at the main evacuation mesh and all door flow evacuation
meshes and this is good. But if you have overlapping (or touching) main
evacuation meshes and you need MESH_ID to specify to which mesh
things go (like DOOR, ENTR, OBST, etc), then you might need two OBST
lines for one door, one for the main evacuation mesh and one for the door flow
mesh, if you do not want the OBST to go also for the other main evacuation mesh.

TimoK

[wnst]

Hi, Timo

 

Thanks for your quick reply.

Yes, I don't have the overlap meshes in my case. Later I will try to do some more complicated cases after I complete this multi-storey case.

I am also interested in FDS6 and will try to compile the latest code. Is there any guide or reference  for FDS6?

 

Thank you!

 

Wnst

 

 

 

2012年6月20日水曜日 16時20分07秒 UTC+9 TimoK:

[TimoK]

Search for EVAC_FDS6 in this discussion forum and you should find:

Topic search results for group:fds-smv AND EVAC_FDS6

Re: FDS+Evac: door selection algorithm By me 9 posts 0 views updated Feb 8 + 1 other

Re: Does it provide leading person that other people will follow him? By Gary_Likai_Hsiao 10 posts 0 views updated 12/23/11 + 2 others

Re: FDS+Evac (FDS 5.5.3 SVN 7031) No Visible or Known Exit Doors By me 5 posts 0 views updated 9/30/11 + 2 others

Re: Visibility conditions on FDS+Evac By me 8 posts 0 views updated 4/21/11 + 1 other

Re: FDS+Evac: Evac Mesh question By Andrea 3 posts 0 views updated 2/7/11 + 1 other

Evacuation version 2.4.1 for beta testing By me 1 post 0 views updated 12/31/10

Re: Evac stars error By me 3 posts 0 views updated 12/2/10 + 2 others

TimoK

[wnst]

Timo

 

Thanks for your information.

I will try later.

 

Wnst

2012年6月20日水曜日 18時03分58秒 UTC+9 TimoK:

[wnst]

Hi, Timo

 

This week I am working on a building with 3 floors upground and 1 floor underground. The staircases have the same projected area.

I successed in the case only with 3 floors upground, and the agents moves from the upper stairs to the exit on the ground floor.

But if the underground floor is added, the agents underground will not use the stairs between B1F to 1F. They seems do not know the stair.

And I  adds a door at the landings on B1F, the agents go directly to the stair in the 2F floor, not the 1F floor.

How to deal with such cases?

 

I also find that some doors in the staircase have not beed diplayed  in SMV.

 

Could you give some advice?

 

Thank you in advance.

 

Wnst

2012年5月22日火曜日 18時49分32秒 UTC+9 wnst:

[TimoK]

Hi!

Well, the usual way to debug your input is to take the "&EVSS" lines away, e.g.,
just delete the "&" characters. Then you will see the agents at the correct main
evacuation mesh z-level, so you know which mesh they are in. The EVSS change
the z-coordinates of the agents so you can not see so easily where the agents are.

Then add "COLOR='some_fds_color_table_color'" to your DOOR/ENTR/EXIT lines so
that different doors show up in different colors in Smokeview. Note that just the front
side of the doors is colored , the back side of a door is some kind of green and the
backside of an entry is cyan (or something like that).

It might be that you can not see all doors etc., if some have same XB. They are on top
of each others. Use Smokeview menu "show/hide", "geometry", "objects". There you
find "evacdoor" and "evacentr". Activate/deactivate them appropriately so that you
just see the ENTRs and just the DOORs. This might help to see where the things
are and are they there where they should be.

And put also "COLOR='xxx'" keywords to your "outflow" VENTs so that you can
see them better in Smokeview. And put DT_PART=0.1 at the DUMP-namelist so
that the time step for the agents is quite short and the agents move more smoothly
in Smokeview so that you can see "the jumps due to door => door connections"
better.

Hope that this helps.

TimoK

[wnst]

Hi, Timo

 

Thank you very much for your advice.

I changed the mesh for the underground stair. And the agents can walk into the stair. But they only walk to the horizontal landing, then go back to the underground.

 

I also add  "COLOR" to some doors. I intended to place some doors on the incline, but they are placed on the upper stair.

For example, as the underground level  is at -4.0, and the 1st floor level is at 0.0, I wanted to place a door on the inclined stair using the following code.

But it turns up to the 1st floor. The position of door will be  changed if it is placed on the inclined stair.  The position change also takes place on the 2nd floor.

&DOOR ID='B1F_StairDoor_in', IOR=-1,
      VENT_FFIELD='1stStairCaseEmesh',
      COLOR='BLACK',EXIT_SIGN=.TRUE.,
      TO_NODE= '1f_StairDoor_in',
      KEEP_XY=.FALSE.,
      XYZ=14.4, 28.8, -3.0,
      XB= 14.2,14.2, 27.8,29.8, -3.1,-2.9,

 

Could you give me more help?

 

Thanks.

 

Wnst

 

 

2012年6月28日木曜日 17時02分50秒 UTC+9 TimoK:

[TimoK]

Well, it might be that you need an evacuation OBST (EVACUATION=.TRUE.) so
that the agents are not dropping to the ground floor.

Well, let's make an example:

Graund floor has: The ground  floor "landing", the stair flight going
from ground floor (z=-4.0, mesh XB=...,-4.1,-3.9, for example) to
an intermediate landing at z=-2.0. And the  intermediate landing
is put to the ground floor evacuation mesh, i.e., XB=....,-4.1,-3.9,
and HEIGHT0=2.0, HEIGHT=2.0 so that it is 2.0m above the
ground floor. The ground floor => intermediate landing stair
flight EVSS is also at the ground floor main evacuation mesh,
so XB=....,-3.9,-4.1, and HEIGHT0=0.0, HEIGHT=2.0 with a
suitable IOR (plus or minus in the ior might make the heights
so that HEIGHT0=2.0, HEIGHT=0.0, i.e., IOR defines your
reference end of your stair flight).

Now the agents at the GF => int.landing + int.landing are belonging
to the ground floor main evacuation mesh.

The landing from int.landing => first floor is put to the first floor main
evacuation mesh, so you have "mesh change" at the place, where
the intermediate landing connects to the stair flight going to the
first door. So, you need here DOOR => ENTR (or DOOR => DOOR)
connection here. The "incoming_door" is at the ground floor main
evacuation mesh, XB=....,-4.1,-3.9, and has IOR pointing towards
the door. The TO_NODE is the name (ID) of the ENTR, that is in
(about) the same (x,y), but at the first floor main evacuation mesh,
so it has XB=...,-0.1,0.1 (if the first floor is at z=0.0m and mesh XB=...,-0.1,0.1).

The stair flight from int.landing => 1st floor is in the first floor main
evacuation mesh: XB=....,-0.1,0.1, HEIGHT0=0.0, HEIGHT=-2.0
(negative z, because flight is below the first floor, of course). And
the IOR so that the reference end of the evss corresponds the
heights given. If not, change IOR or height0 and height.

And now the outflow vents etc:

The "incoming_door" needs an OUTFLOW vent, so make an outflow
vent with the  same XB (i.e., at the ground floor main evacuation mesh)
here. And put there an OBST also (use EVACUATION=.TRUE. so that
this does not go to the fire simulation). VENT needs at least one grid
cell thick obst behind it. And the evacuation mesh needs also an obst
here so that there is no way for the agents to cross the door line and
"go back to the ground floor mesh". Well, the agents would not do
this, if the DOOR is nicely defined, but the guiding flow fields for
evacuation are not good, if you do not have here an obst. The staircase
should be "a dead end" when just the parts in the ground floor mesh are
seen.

And for the same reason, put on evacuation OBST behind the ENTR
that is in the first floor mesh. This way this is also "a dead end" in
the first floor mesh, so that the guiding flow fields for evacuation are
nice in the first floor mesh. The main evacuation mesh should contain
one "sealed" part, where all agents and all doors (and especially
all outflow vents) are.

TimoK

[wnst]

Hi, Timo
Thank you very much for your detail instruction.
I changed the mesh of int.landing=>1st floor. The "incoming door" is placed at the int.landing, but the OBST is on the ground floor, not on the landing. And as to setting the door behind the "incoming door", it jumps to the stair flight above. So the agents go directly to the upper stair , then they go to the 1st floor. In my case, the agents above 1st floor go down, while those in ground floor go up, they merge in 1st floor.
How to deal with these problems?

Thanks in advance.

Wnst

[TimoK]

Well, you should decide how you devide your stair landings and flight to
three floors: GF, 1stF, 2ndF. The final exit is at 1stF.

So, if the stair geometry is same in all floors then you can have at most
two flights in one floor (one floor = one main evacuation mesh).

So, put next things to the 1stF mesh:

    flight 1stF => landing between 1stF and GF
    flight 1stT => landing between 1stF and 2ndF

It depends on your case, if you should add either of the two
intermediate landings to the 1stF mesh. It depends how you
have defined your upper part of the stairs. If the landing between
1stF and 2ndF belongs to the 1stF mesh, then you should put
the landing between 1stF and GF to the GF mesh. You can not
have two things with the same (x,y) in the same mesh.

So, if your 1stF <=> 2ndF landing is at the 1stF mesh, then you
have to choose:

 Put an evss representing the fligth 1stF<=>1stF/2ndF intermediate landing
 to the 1stF mesh.
 
Put an evss representing the flight 1sfT<=>1stF/GF landing to the 1stF mesh. 
Put a door=>entr at the bottom of the flight from 1stF to the intermediate
landing 1stF/GF. And place OBSTs there also at 1stF and GF floor
meshes.

Put the landing 1stF/GF to the GF mesh and put the landing=>GF flight
also to the GF mesh.

TimoK

 

[wnst]

Hi, Timo

Many thanks for your help.

So far I have not succeeded in my case, I found I made some mistakes in setting the mesh  for lands and flight.

As you told me about the meshes for 3-storey building on May 25, I am working on a building with more than 4 storeys.

For example, it is a 4-storey building,

 

 GF => flight1 => landingA =>flight2 => 1stF

1stF <= flight3 <= landingB <= flight4 <= 2ndF

 2ndF <= flight5<= landingC <= flight6<= 3rdF

 And DOOR => ENTR should be added on the landings.

 

GF => flight1 => landingA =>DOOR_A => ENTR_A=>flight2 => 1stF

1stF <= flight3 <= landingB<=  ENTR_B<=DOOR_B <= flight4 <= 2ndF

2ndF <= flight5<= landingC<= ENTR_C<= DOOR_C <= flight6<= 3rdF

Therefore the mesh can be defined as follows:
(1)GF mesh: flight1, landingA, DOOR_A => ENTR_A(at the 1st floor mesh)

(2)1stF mesh: ENTR_A, flight2 ,ENTR_B, flight3

 (3)2ndF mesh: flight5, ENTR_C, flight4, landingB, DOOR_B => ENTR_B(at the 1st floor mesh)

 

(4) 3rdF mesh: flight6, landingC, DOOR_C => ENTR_C(at the 2nd floor mesh)

  How about the mesh of flight5. ENTR_C?  Can they be put on the 2ndF mesh?

 

 And  XBs for DOOR_A  and  ENTR_A are the same? In fact the 1st Mesh has the level of Z =0.9-1.1, while the GF mesh is on the Z=-3.1,-2.9. (As the floor height is 4m in my case.)

Am I right?

I am confused that sometime the ENTR or DOOR will be not displayed in SMV.

 

Thanks in advance.

 

Wnst

 

2012年7月3日火曜日 23時00分51秒 UTC+9 TimoK:

[TimoK]

Well. lets see. You should keep in mind that the building and the stairs
are represented as two-dimensional horizontal planes in FDS+Evac. And
the planes are defined by the main evacuation  meshes. The stairs and
landings must belong to one and only one main evacuation mesh. So,
you should first pick as many plain A4-papers as you have floors in
your building (the number of your main evacuation  meshes usually).
You should draw a layout to each A4-papers so that a specific
stair flight is just in one A4-paper. And same for the landings.
Now you should know to which main evacuation mesh each flight
and landing belongs. And you see from the layouts where the agents
should jump from one A4-paper to the next A4-paper. And at this
point you should put door => entr. And at this point there should
be a solid wall at both A4-papers, so put some EVACUATION=.TRUE.
OBSTs at the both A4-papers if there are no walls already there.
And make sure that the stair flights (and landings) have "railings",
so that there is walls in the A4-papers at the sides of the flights.
The EVSS should mave MESH_ID (well, this is an optional parameter)
and XB that fits the corresponding A4-paper (i.e., the main evacuation
mesh).

If you have places, where there would be two stair flights (or landings)
on top of each other in some A4-paper layout, then you know that
you have a problem. Then you should think how you can draw the
geometry to A4-papers so that you will not have things on top
of each others at a single A4-paper. Each (x,y) point at the
A4-paper can just have one z-coordinate. It is just as you would
have a pair of scissors (or a knife) and you could cut the pieces
out of the A4-paper and just translate them in z-direction. You
should be able to construct you geometry this way. And each
of your A4-paper is different color, so you know which piece of
paper (cutted out using the scissors) belong to which A4-paper.

In some cases your stair (etc) geometry could be such that you
can not put it to the A4-papers as mentioned above. Then  you
should define a new main evacuation mesh that is between you
two floors. This mesh could have different XB and IJK than the
others. It could just have XB that is about the size of your stairs
(size means (x,y) size, the z should be somewhere between the
two "old" meshes). Now you have an additional A4-paper that
is between the two "old" ones and you should be able to
draw the two-dimensional layouts so that there are not many
overlapping parts with different z-coordinate on the same
A4-paper.

And the stairs should be "dead ends" on the A4-papers, where the
"dead-end" is at the location, where you have door=>entr connections.
The dead-ends could have outflow vents, depending on the movement
direction of the agents. In your case:

GF => door(at GF) => entr(at 1stF) => 1stF (well, might be a fligth
or a landing that belongs to 1stF, but anyhow the agents are in the
1stF mesh). Here you need an "outflow" vent at the door that is
in the GF, so that the agents can find a way to this door. And you
need this "outflow vent" at the main evacuation mesh of the GF and
if you use the additional door flow evacuation meshes (i.e., using the
door selection algorithm and you have more doors than just this
door at the stairs at the GF) then you should have this vent also
in the additional door flow evacuation mesh of this door.
The TO_NODE is the ID of entr(at 1stF). And the VENT_FFIELD of
the door(at GF) should be the ID of the door flow evacuation mesh,
if you have that (or if no additional meshes then the GF mesh ID).

I can not explain this better. Think everything as two-dimensional
planes, do not think that the stair flights and landings are at
their correct positions. Place them to the two-dimensional
horizontal planes. And then the transformation between the
different two-dimensional horizonal planes are done like
in the famous series: "Beam me up, Scotty" -style using
the door=>entr as a "transporter".
(http://en.wikipedia.org/wiki/Beam_me_up,_Scotty,
http://en.wikipedia.org/wiki/Transporter_%28Star_Trek%29)

TimoK

[wnst]

Hi, Timo

Thank you very much for your detail instrcution.

Owing to your illustration of the A4-Paper for one stair floor, I learned more about the EVSS and evaluation mesh. Thank you.

 

The geometry of my case is not so complicated, and it is common in buildings.

So I simplified my case to a 3-storey building.

I successed in the simulation for the agents  in the B1F (the lowest floor), they can go up to the 1F. But the agents in the 2F(second floor) can not go downstairs.

What  I have done in the 2F is the same with that in B1F, why the results are different?

Maybe I have not caught the meaning of " Dead End " .

 

I am hesitated to aks you if you could help to check my fds file?

 

Best regards,

 

Wnst

 

 

2012年7月5日木曜日 17時06分03秒 UTC+9 TimoK:

[wnst]

Hi, Timo

 

Hi, the simulation seems to be successful  in the 3-storey case.

The only problem is two doors belonging to the 1st floor are dispeared, which looks not so nicely in SMV.

 

I am trying to find the problem.

 

Thanks,

 

Wnst

2012年7月6日金曜日 12時04分06秒 UTC+9 wnst:

[TimoK]

Nice that you got it more or less working.

The missing doors in SMV: It might be that there are some doors
and entries in the Smokeview picture at the same place, so that
you can not see them all. You could tru activate/deactivat the
showing of ENTRs and DOORs using Smokeview menu "show/hide"
and there submenu geometry and there submenu objects and
toggle "evacdoor" and "evacentr" to show doors+exits and entries.
If the XB of a door/exit/entr matches an EVSS at the same main
evacuation mesh, then the door/exit/entr is displayed in the
Smokeview using the EVSS data, so it is not at the main
evacuation mesh plane anymore.

See the attached png-image, it is the stair_example.fds case at
the FDS+Evac homepage (Examples subpage and the at the
bottom you find "More examples can be found here.", where
the word "here" is a link to: http://virtual.vtt.fi/virtual/proj6/fdsevac/examples_fds5.html)

This stair has the upper landing as a part of the upper mesh, so there is
a door with an outflow vent there. And there should also be an obst, but
in this case the door and the vent are at the outer boundary of the mesh,
so no need for the obst, the boundary is acting like an obst. This door
has TO_NODE= the ENTR that is at the top of the upper stair flight, but
now this flight is belonging to the lower mesh, so here you have the
agents the change upper mesh => lower mesh. So, the agents see
right after the door=>entr the lower mesh geometry. The evacuation
mesh geometries are shown as magenta outlines (yellowish color is
for the "true geometry", i.e., the fire mesh geometry). You can see
that at the lower mesh, there is an obst at the entr position, so there
is a "dead-end" here. The "dead-end" obst is shown at the lower
mesh level, the entr is at an EVSS so it is shown a little bit higher,
in this case about 3m higher (at the level of the upper mesh, because
the EVSS has height and height0 given so). So, the agents jumping out
of the entr have no other directions than go towards left, then at the
middle landing (x,y) they have to turn and then continue towards
the hole that connects the staircse to the lower floor.

Note that there can not be any agents "under the upper stair flight",
so you need the OBST at the position of the ENTR at the lower mesh
to close this path for the lower floor agents. If you would need agents
"under the upper stair flight" then you should change the strategy:
Put the upper flight to the upper mesh and put the door=>entr at
the connection between the upper flight and the intermediate landing.
And if this is not enough (i.e., you need agents under the intermediate
landing at the lower mesh) then yous should also put the intermediate
landing to the upper mesh and the door=>entr should be at the
connection between int.landing and the lower flight. And so one.
It is up to modeller, how to put different flights and landing to
the different meshes, i.e., which part of the staircase is
projected to which mesh (mesh = A4 paper).

TimoK

[wnst]

Dear Timo

 

Thank you very much for your detail instruction.

I think I understand the meaning of "Dead-End" now.

 As the EVSS is the connection between the upper floor and the lower one, there is an  inlet (VENT) for agents at the entrance of the EVSS.

And there should also be an OBST at the outlet of the EVSS.

 

"mesh = A4 paper" is very good for understanding. Thanks.

 

In order to see the whole building for evacuation, I set the fire mesh from the Basefloor to the sencond floor and turn off  SLCF PBZ=-3.0 or SLCF PBZ=1.0.

(& is deleted for SLCF PBZ=-3.0,....).  But the whole building was not shown out. The results are the same with the former.

 

I want to ask if it is possible to output the position of agents during evaculation.

 

Also, as the evacuation speed  for Adult is set to be 1.25+(-)0.3 m/s , if it is possible to change this value to be  a constant. I would like to do some comparison with the Route B Method in Japan.

 

Thanks in advance.

 

Best Regards

 

Wnst

 

 

 

2012年7月6日金曜日 17時47分51秒 UTC+9 TimoK:

[TimoK]

I was on a summer holiday without any internet, so sorry for a
late answer. First the easy answers:

1) How to set a constant speed:

   PERS namelist and there the keyword VELOCITY_DIST. The manual has
  "The type index of the unimpeded walking speed distribution, see Table 4."
   The VELOCITY_DIST=0 will use a constant speed for all agents that are
    generated utilizing this PERS information. The constant speed is given
    at the VEL_MEAN parameter, so set VEL_MEAN=1.25 (or whatever speed
    you need).

2) "But the whole building was not shown out." Well I do not understand this
    question. Is your fire mesh XB large enough? If it is large enough, try to
     take the "Exterior surfaces" of in Smokeview. Use the "Hide/Show" menu
     and there "Geometry" => "Surfaces" and toggle with the "Exterior". The
     boundaries of the evacuation meshes might be blocking the view.

3) "I want to ask if it is possible to output the position of agents during evaculation."
    Well, you can see the agents  during the simulation. They should be visible
    when you do a fire+evacuation calculation like in a plain evacuation calculation.
    If you want to see the results of a plain evacuation calculation later using the
    fire geometry, then you should play around with the .smv files and/or copying
    the .prt5 files. There are some threads on this discussion group how to do this.
    Search for these.

    The .prt5 files of the main evacuation meshes (there is a mesh index in the .prt5
    file names) contain the x,y,z information for each agent in the calculation as
    a function of time. The time steps are dictated by the DT_PART keyword on the
    DUMP namelist. There is some default DT for prt5 files, see the FDS guide for this.

 TimoK

[wnst]

 Welcome, you are back.

I heard most Europeans would spent 2 months on vacation in summer, and you only spent 1 month. 

Thank you very much for your answer. It is very helpful to compare  Evac and the Route B method, the performance-based fire safety verification method in Japan.

But I still have some questions I will ask you later.

Thank you.

Lei

2012年8月6日月曜日 22時41分31秒 UTC+9 TimoK:

[wnst]

Hi, Timok

I would like to simulate the tsunami evacuation, which means the agents go to the rooms from the outside.

As to the namelists DOOR, EXIT and ENTR, they have the properties, like COUNT_ONLY, EXIT_SIGN,TIME_CLOSE. Is it possible to simulate the following case .

When the agent number reaches a given number (for example,100),  then the DOOR or Exit is closed.

Could you please give me some more help?

 
Best Regards,

Lei

P.S.

Wnst is my nick name for gmail.

Perhaps I got the wrong information about the 2-month vacation in Europe. Most academics have to spent more time on their works.

2012年10月1日月曜日 17時43分11秒 UTC+9 wnst:

[TimoK]

Dear Lei,

You can not close a door nor an exit when there is some amount of persons
gone through it. But if you do not care too much what is happening inside
the rooms, then you can use DOOR => CORR => "dummy EXIT".

The CORR has a parameter MAX_HUMANS_INSIDE. So, set this to 100
and set also EFF_LENGTH and FAC_SPEED so that the agents are
walking a really long time in the CORR, so they do not reach during
the simulation the "dummy EXIT". It might be that you could also
set the "dummy EXIT" as closed. But check this if this is working or
not. I do not remember if the CORR routine is checking the status
of the EXIT/DOOR.

TimoK

[Lei]

Dear Timo,

Thank you very much for your advice.

I will try to use the CORR in my simulation and tell you the result later.

Many thanks.

Sincerely yours,

Lei

2012年10月22日月曜日 22時01分01秒 UTC+9 TimoK: