Re: [fds-smv] Heat transport: conduction and convection

586 views
Skip to first unread message
Message has been deleted

Randy McDermott

unread,
Aug 25, 2014, 9:12:21 AM8/25/14
to FDS and Smokeview Discussions
If this is copyrighted material, please do not post to the forum, just post the citation or link to the journal.

Read the user guide, you can specify the heat transfer coefficient on the SURF line.  Section 8.2.2.


On Mon, Aug 25, 2014 at 8:42 AM, buha <zeno.g...@gmail.com> wrote:
Hello!

I'm trying to replicate a numerical simulation, performed by a professor from my faculty, using FDS.

He analyzed a wall made of 2.5 cm plaster + 10 cm mineral wool + 2.5 cm plaster using the finite element method. On one side of the wall he defined the parametric temperature-time curve from Eurocode 1 Part 1-2. On the other side he monitored the obtained temperature and also the temperatures inside the wall. I attach the paper.

What puzzles me is how to define the convection coefficients used in the paper: 25W/(m2*K) for the fired surface and 8W/(m2*K) for the unfired surface.

Thank you!

--
You received this message because you are subscribed to the Google Groups "FDS and Smokeview Discussions" group.
To unsubscribe from this group and stop receiving emails from it, send an email to fds-smv+u...@googlegroups.com.
To post to this group, send email to fds...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/fds-smv/140a3461-1a1d-4b37-ac3f-d8f7c8558161%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

buha

unread,
Aug 25, 2014, 9:21:45 AM8/25/14
to fds...@googlegroups.com
Thanks for the answer. I will read Section 8.2.2.

Sorry for the attachment. The link to the journal is: http://www.intersections.ro/archive/2005/No1/No1_eng.pdf

page 59

title: Procedure for checking the requirement referring to fire insulation capacity of a non-structural wall from the fire compartiment 

authors: Dan Diaconu-Sotropa, Mihaela Ibanescu

buha

unread,
Aug 25, 2014, 12:32:04 PM8/25/14
to fds...@googlegroups.com
Because I violated the posting rules I rewrite the first message because I can't get the same results.

I'm trying to replicate a numerical simulation, performed by a professor from my faculty, using FDS.

He analyzed a wall made of 2.5 cm plaster + 10 cm mineral wool + 2.5 cm plaster using the finite element method. On one side of the wall he defined the parametric temperature-time curve from Eurocode 1 Part 1-2. On the other side he monitored the obtained temperature and also the temperatures inside the wall. See the articole above!

The initial temperature is 15*C and I don't have radiation.

So far I defined two surfaces:
- a layered surface with the material properties from the article;
- heater with the parametric temperature-time curve;
Thanks to Randy McDermontt I assigned the surface convection coefficients.

I defined an obstacle wich have the two surfaces on oposite sides. 

I got slightly different temperatures than the temperature from the article (it is ok for me) but the inside temperature are unrealistic. I missed something in the model, please take a look.
wall.fds
temperatures.png

dr_jfloyd

unread,
Aug 25, 2014, 12:40:22 PM8/25/14
to fds...@googlegroups.com

Kevin

unread,
Aug 25, 2014, 12:56:44 PM8/25/14
to fds...@googlegroups.com
I do not think the case has been properly set up. You have a single obstruction:

&OBST XB=0.0,0.45,0.0,0.45,0.15,0.3, SURF_IDS='wall','INERT','heater'/ Obstruction

One side of the obstruction is the layered material, the other is a "heater" consisting of a ramped surface temperature. This is not the way one would do a heat conduction calculation in FDS. Take a look at some of the example cases in Heat_Transfer, in particular heat_conduction_a.fds, heat_conduction_b.fds, etc.

buha

unread,
Aug 26, 2014, 4:59:01 AM8/26/14
to fds...@googlegroups.com
Thank you Kevin!

I modified the example you suggested and the temperatures are very close to those from the article. 

I noticed a little problem:
1. if a use 3 layers (2.5cm plaster + 10cm mineral wool + 2.5cm cm plaster) I don't have a linear variation of temperature in the mineral wool;
2. if a use 4 layers (2.5cm plaster + 5cm mineral wool + 5cm mineral wool + 2.5 cm cm plaster) the variation of temperature in the mineral wool is almost liniar;
temperatures.png
heat_conduction_a_mod.fds

Kevin

unread,
Aug 26, 2014, 9:19:21 AM8/26/14
to fds...@googlegroups.com
Add this line to your input file:

&PROF ID='prof', QUANTITY='TEMPERATURE', XYZ=0.0,0.0,0.0, IOR=3, FORMAT_INDEX=2 /

This will give you the temperature profile through the wall. It will also show you the location of the numerical grid within the wall. Also, get rid of that RAMP function. Use a few RAMP lines to approximate it.

buha

unread,
Aug 27, 2014, 5:13:39 AM8/27/14
to fds...@googlegroups.com
Thank you, Kevin!

Generally speaking, if I approximate a RAMP function the simulation will run faster?

Kevin

unread,
Aug 27, 2014, 9:13:41 AM8/27/14
to fds...@googlegroups.com
No, the case won't run faster, but it will certainly be easier for others to understand if you use less RAMP lines. The best way to get help fast is to make the case as simple as possible. 10000 RAMP lines is not simple.

buha

unread,
Sep 30, 2014, 8:03:25 AM9/30/14
to fds...@googlegroups.com
Hello again!

I return to this topic with the next question: It is possibile to define two convection coefficients on the analyzed wall? 
According to article, form the second messenge, it was used:
for the fired surface 25 W/(m^2*K)
for the unfired surface 8 W/(m^2*K)

I'm asking this because I modeled the same case in Ansys (using the mechanical solver).

I'm sorry for my English!

Randy McDermott

unread,
Sep 30, 2014, 8:07:04 AM9/30/14
to FDS and Smokeview Discussions
User Guide Sec. 8.2.2, Specifying Convective Heat Transfer Coefficient.

--
You received this message because you are subscribed to the Google Groups "FDS and Smokeview Discussions" group.
To unsubscribe from this group and stop receiving emails from it, send an email to fds-smv+u...@googlegroups.com.
To post to this group, send email to fds...@googlegroups.com.

buha

unread,
Sep 30, 2014, 12:40:05 PM9/30/14
to fds...@googlegroups.com
I defined an obstruction having two opposite sides made of same materials with the same thickness.
On the fired surface I defined HEAT_TRANSFER_COEFFICIENT=25 and on the opposite surface I defined HEAT_TRANSFER_COEFFICIENT=8.

To simplify to input file, I didn't use a RAMP function, the ambient gas temperature is constant (1200 C).

Whatever heat transfer coefficient I use for opposite surface the temperature is the same.


ANSYS: 
fired surface: 1882 C
opposite surface: 64 C

FDS:
fired surface: 1882 C
opposite surface: 34 C

What I did wrong? 
340_min_eng.fds
ansys.jpg

Julio Cesar Silva

unread,
Sep 30, 2014, 12:49:19 PM9/30/14
to fds...@googlegroups.com
Could you explain which variables are you transferring from FDS to Ansys?
Also, are you considering a 1D or 3D consuction model at Ansys?

Regards, Julio


For more options, visit https://groups.google.com/d/optout.



--
___________________________________
Julio Cesar Goncalves da Silva, D.Sc.

Guest Researcher, Fire Research Division
National Institute of Standards and Technology
Tel: +1 301-975-6577
Mobile: +1 240-478-2850

dr_jfloyd

unread,
Sep 30, 2014, 12:56:51 PM9/30/14
to fds...@googlegroups.com
SURF defines the entire surface from the front of the obstruction to the back of the obstruction (read the users guide chapter on heat transfer in solids).  The backside temperature is only used if the OBST it is applied to is equal to or less than once cell thick and has the BACKING set to EXPOSED.   We don't currently have an input to allow you to set different front and backside heat transfer coefficients for the same SURF input; however, seems like a reasonable thing to add for a future release.

buha

unread,
Sep 30, 2014, 1:37:13 PM9/30/14
to fds...@googlegroups.com
I'm not transferring anything from FDS to Ansys, I'm just model the same problem in two different software: applying a temperature on a wall (made from 2.5 cm plaster + 10 cm mineral wool + 2.5 cm plaster) and I want to get the  temperatures on the wall's surfaces.

buha

unread,
Sep 30, 2014, 1:39:01 PM9/30/14
to fds...@googlegroups.com
Thanks for the reply, dr_jfloyd!

buha

unread,
Sep 30, 2014, 1:39:55 PM9/30/14
to fds...@googlegroups.com
I'm not transferring anything from FDS to Ansys, I'm just model the same problem in two different software: applying a temperature on a wall (made from 2.5 cm plaster + 10 cm mineral wool + 2.5 cm plaster) and I want to get the  temperatures on the wall's surfaces.

marți, 30 septembrie 2014, 19:49:19 UTC+3, jcsilva a scris:

TimoK

unread,
Oct 2, 2014, 5:11:16 AM10/2/14
to fds...@googlegroups.com
So, you could try:

 1) Make your wall as one grid cell thick. This way you can use BACKING='EXPOSED'.

 2) Define two SURF-lines. One for the fire side and one for the ambient side.
     Your wall seems to be symmetrical, so the two SURF lines do not differ
     much, they just have different HEAT_TRANSFER_COEFFICIENT.

 3) Note, you should have some grid cells on the ambient side. I.e., you wall OBST
     can not touch the outer boundary of your simulation geometry (the MESH boundary).
    So, add at least one grid cell thick air space before the mesh boundary on the ambient
    side and set the temperature here (INIT) to the ambient value. Add enough gid cells to the
    ambient side, if you want to model nicely the convection there. But if you need just HTC=8 with
    ambient temperature gas, then you could make the ambient side so that you blow there ambient
    temperature gas, say, below, and have open boundary at top, and solid boundaries at side. This
    way your ambinet side air do not heat up (well, it heats up, but new cool air is put there) and
    the heat lost of your wall (by convection) is Qconv_lost = Convective_coeff * (T_back_wall - T_amb),
    at least approximately. The radiation transfer is to ambient, if you have ambient boundaries at the
    non-fire side.

TimoK



Reply all
Reply to author
Forward
0 new messages