flame spread on thin paper

[gforney]

Please complete the following lines...

Application Version: FDS 5.4.2
SVN Revision Number: 4957
Compile Date:
Operating System: windows

Describe details of the issue below:

  Hi,
I am a new user of FDS.I am trying to simulate flame spread on thin paper 
(thickness 0.08mm) under microgravity in a long horizontal channel. The 
gas combustion was single step finite rate reaction and the solid phase 
adopted the FDS' sample charring_solid.fds. The parameters were from the 
article published before. But the reaction temperature was low and lots of 
parameters can't be confirmed. Because we didn't know the actual pyrolysis 
fuel,the molecular weight of the fuel was assumed and the BOF, E, NU, N_S 
and HEAT_OF_COMBUSTION was assumed either.

When the horizontal channel is low enough the flame can spread, but it 
will extinguish when the horizontal channel increase gradually. So I don't 
know what should I do to get better result? We have done the drop 
experiments before so we have the data about it. 

I would be very grateful if anyone could provide me some advice on this.
Thanks very much,

rentan

Original issue reported on code.google.com by rentanx@yahoo.cn on 2010-04-05 14:22:57

---

• Attachment: cell-h10-10cms.fds

[gforney]

Is there something in FDS that you believe is not working properly? If so, this is 
the right place to submit a bug report. Otherwise, you can start a general 
discussion within the Discussion Group. However, it is not a good idea to submit an

input file and ask people to comment on it. You need to do a better job of 
identifying the important physics or parameters of the problem. Saying that you 
don't know what to do to get a "better" result is not helpful. If you do not know 
the properties of the fuel, and you are doing a DNS calculation, what do you mean by

a "better" result?

Original issue reported on code.google.com by mcgratta on 2010-04-05 15:12:54

[gforney]

    I think the gas phase combustion model is not working properly now. I found the

flame speed changed a lot when I changed the molecular weight of the fuel, BOF, E,

NU, N_S and HEAT_OF_COMBUSTION. Compared with the results of the drop test the flame

speed in different flow speed was not very well. Do you think it is proper to use 
DNS calculation when I am not sure the properties of the fuel?  And then how to set

it more properly?
Thanks,

Original issue reported on code.google.com by rentanx@yahoo.cn on 2010-04-05 15:59:27

[gforney]

Why do you expect the flame speed not to change when you change all of these 
properties? What do you think is wrong with the combustion model?

Original issue reported on code.google.com by mcgratta on 2010-04-05 16:14:27

[gforney]

The flame speed is about 1 cm/s when flame spread in the long horizontal channel 
with liwer height such as 1~3 cm. When I adjusted the HEAT_OF_COMBUSTION to a bigger

one 35000 Kj/Kg which was from the article 'Enclosure Effects on Flame Spread Over

Solid Fuels in Microgravity' published in 2002 wrote by Yuji and you, the flame 
speed was 4 cm/s. At the same time you cosidered the co product so I don't konw how

to use the combustion model when there are lots of parameters can't be confirmed.

Original issue reported on code.google.com by rentanx@yahoo.cn on 2010-04-06 00:49:01

[gforney]

Different model, different assumptions, different parameters. I think you need to 
look at each and every parameter, plus the model assumptions and formulate your own

description. It's a pointless exercise to expect similar results when there are so

many uncertain parameters that go into these flame spread calculations.

Original issue reported on code.google.com by mcgratta on 2010-04-06 15:32:53

[gforney]

Thanks for your advice! Some questions want to ask you. Is NU the stoichiometry of 
the chemical reaction for each SPEC based mass or molecular volume? When the oxygen

supply changed caused by the gas flow or channel height, will the NU changed either?

Also the products will changed, CO will decrease as oxygen increases. How to set it

using DNS model? And the soot will produce as oxygen increases, so the solid phase

will adjust as oxygen changes. Different models should be used to simulate flame 
spread process. Am I right?

Original issue reported on code.google.com by rentanx@yahoo.cn on 2010-04-07 05:33:42

[gforney]

Please read the FDS Technical Reference Guide and then ask specific questions about

the various algorithms. I cannot spend hours answering your questions when they are

all addressed in the Guide. 

Original issue reported on code.google.com by mcgratta on 2010-04-07 12:53:18

[gforney]

Thanks for your advice! James S T’ien simulated the flame spread process in 1979 in

the article ‘A theory of flame spread over a solid fuel including finite-rate 
chemical kinetics’ which he assumed the PYROLYZATE was C6H10O5. I tried to use the

simplified model to simulate the process.  But there are one problem when I simulate

the half height 5 mm,the flame speed decreased when the gas velocity increased but

the fire temperature is not changed. But our drop tests and others before are that

the flame speed increased first with the gas flow increased and then decreased. 
I would be very grateful for your advice on this.

Original issue reported on code.google.com by rentanx@yahoo.cn on 2010-04-14 13:00:01

---

• Attachment: tien-half-5mm.opj
• Attachment: zhi-half-5mm-8cms.fds

[gforney]

I have no advice to give. I have not performed calculations like this for about 10 
years. 

Original issue reported on code.google.com by mcgratta on 2010-04-14 14:27:12

[gforney]

There is another question. When I use finite-rate combustion model, the NU is the 
stoichiometry of the chermical reaction. Does the NU change with the gas flow varies

which it means the real gas mole ratio or ideal stoichiometry ratio?  Thank you!

Original issue reported on code.google.com by rentanx@yahoo.cn on 2010-04-14 14:49:07

[gforney]

I do not understand what you are asking. With the finite-rate model, you specify the

reaction stoichiometry. The NU values do not change during the calculation.

Original issue reported on code.google.com by mcgratta on 2010-04-14 15:15:59

[gforney]

There is another question. I simulate the flame spread with the thickness is 0.04 mm

which the mesh grid is 0.1 mm although I use the  CELL_SIZE_FACTOR  = 0.25. But the

output file solid phase nodes just have two.
 Solid Phase Nodes (m):      
            0   0.00000
            1   0.00004
when I decrease the mesh grid to 0.05 or more small, the result is the same. So 
maybe there are error. Thank you!

Original issue reported on code.google.com by rentanx@yahoo.cn on 2010-04-15 15:03:00

---

• Attachment: zhi-h5-z50-8cms.fds

[gforney]

I do not believe this is an error. There is an algorithm that determines the cell 
size, and it has determined that .00004 m is sufficient to accurately compute the 
heat transfer. It is essentially thermally thin.

Original issue reported on code.google.com by mcgratta on 2010-04-15 16:33:47

[gforney]

Thanks for your kindhearted advice. There is another questions about the upper input

file. The mesh grid in X direction is 1mm while the mesh grid in Z direction is 
0.1mm in order to save time. Is it proper? 
I try to use diffent mesh grid when the mesh grid in X direction is 0.5mm, the 
result (flame speed)don't change even the grid is 0.1mm in X direction. 
I would be very happy for your advice on this.

Original issue reported on code.google.com by rentanx@yahoo.cn on 2010-04-19 15:22:15

[gforney]

I can only say that the discretization error is second order in the cell size. This

means that the error associated with 1 mm cells may cancel out accuracy gained with

0.1 mm. But that is your decision, and you need to justify with a grid resolution 
study.

Original issue reported on code.google.com by mcgratta on 2010-04-19 15:44:19

[gforney]

There is another question. I assumed the fuel mole weight is 162 and the simulate 
the flame spread process in full scale channel with the thin paper in the middle of

the channel. But the flame shape was unsymmetrical which was contrary with the 
normal concept. 
I would be very happy for your advice on this.

Original issue reported on code.google.com by rentanx@yahoo.cn on 2010-04-29 15:53:26

---

• Attachment: 14mm-0g-20cms.jpg
• Attachment: zhi-h10--mw-162-5cms.fds

[gforney]

Gravity? Atmospheric stratification? I don't know exactly what it could be. If you 
think there is a bug in the software, create a very simple case that demonstrates it.

Original issue reported on code.google.com by mcgratta on 2010-04-29 16:05:00

[gforney]

I set the gravity as  
GVEC=0.0,0.0,0.0 because I want to simulate the microgravity condition. I set the 
atmospheric stratification as 
&SURF ID ='SUPPLY', VEL=-0.05, MASS_FRACTION(2) = 0.23,COLOR='BLUE'/ which means the

gas flow is 5cm/s and O2 concentration is 0.23. Is it not proper?
I assume the fuel is C6H10O5 considering the ACTIVE produciton which was alike with

the example in fds.
At the same time I tried the simple model which the solid phase reaction is single

step reaction. The flame shape is symmetrical in microgravity condition. The 
difference between the two model is the solid phase reaction. 

Original issue reported on code.google.com by rentanx@yahoo.cn on 2010-05-08 14:57:43

[gforney]

In your input file, it appears that the heating element is not symmetric. Are you 
sure that the calculation is set up to be symmetric?

Original issue reported on code.google.com by mcgratta on 2010-05-09 20:07:36

[gforney]

I tried to use small mole weight fuel in FDS file as in the attached file. I 
simulate the flame spread in microgravity conditon and the flame shape is symetric

but the heating element is not changed. So I think maybe the role of the heating 
element is not important.
I would be very happy for your advice.

Original issue reported on code.google.com by rentanx@yahoo.cn on 2010-05-10 00:45:00

---

• Attachment: zhi-0g-h14-rou400-x200-z175-3.5cms-t800-1.0s.fds
• Attachment: symmetric.jpg

[gforney]

Your paper is zero grid cells thick. This means that there is only one normal 
component of velocity at the paper surface. In order to have symmetry, you need to

make the paper sample one cell thick so that there can be normal components of 
velocity in opposite directions.

Original issue reported on code.google.com by mcgratta on 2010-05-10 12:03:39

[gforney]

Thanks for your kindhearted advice. I compute some case again which the computation

domain height is 10mm and the z direction mesh grid I adopt is 50, 100, 125. When 
the mesh grid is 125 which make the paper sample one cell thick the flame shape is

symmetrical.

And there were many people researching the gas phase radiative and solid surface 
radiation in flame spread process. How to get it in fds model? I think the HRR 
indicates the flame hoc and radiative loss to the boundries indicates the gas phase

radiative and the net heat flux at the channel wall indicates the wall heat flux.Am

I right?
Then how to get the solid phase surface radiation? 
I would be very happy for your advice.

Original issue reported on code.google.com by rentanx@yahoo.cn on 2010-05-11 00:34:37

[gforney]

&BNDF QUANTITY='RADIATIVE HEAT FLUX' /

Original issue reported on code.google.com by mcgratta on 2010-05-11 12:16:28

[gforney]

There is another question. I read the hrr.csv file and find the sum of the RAD_LOSS,

CONV_LOSS and COND_LOSS is larger than the HRR as in attached file. It makes energy

is not banlace. Maybe there is some error. And the unit of RADIATIVE HEAT FLUX and

NET HEAT FLUX is KW/M2 but I simulate the two dimension model. I don't konw how to

connect it with the HRR which the unit is KW. I think the unit should be KW/M. Am I

right?

Original issue reported on code.google.com by rentanx@yahoo.cn on 2010-05-15 15:36:43

---

• Attachment: h10-0g-15cm.opj

[gforney]

Typically, HRR=CONV_LOSS+COND_LOSS

Radiation contributes to COND_LOSS.

Original issue reported on code.google.com by mcgratta on 2010-05-15 19:13:58

[gforney]

Also I try the sum of CONV_LOSS+COND_LOSS, and the sum is larger than HRR as 
attached file upper. I don't konw how to connect RADIATIVE HEAT FLUX with the HRR 
because the units are different. 
I would be very happy for your advice.

Original issue reported on code.google.com by rentanx@yahoo.cn on 2010-05-16 01:18:27

[gforney]

I cannot open an opj file.

Original issue reported on code.google.com by mcgratta on 2010-05-17 21:32:55

[gforney]

Now I change it to pictures, you can see it. 
I would be very happy for your advice.

Original issue reported on code.google.com by rentanx@yahoo.cn on 2010-05-18 14:34:24

---

• Attachment: Graph1.png
  
• Attachment: Graph2.png
  

[gforney]

The difference might be due to the very large value of width of the domain. You 
should set Y2-Y1 on the MESH line  (XB=X1,X2,Y1,Y2,Z1,Z2) to something like 1 mm.

Original issue reported on code.google.com by mcgratta on 2010-05-18 15:20:17

[gforney]

Thanks for your advice. Today I try a simpler case in attached file and use 0.1m and

0.001m width of the domain. The results are in attached file. Maybe it is not the 
key. I have seen some other questions about this at open issues but it seems not to

be relative. 

Original issue reported on code.google.com by rentanx@yahoo.cn on 2010-05-19 15:42:36

---

• Attachment: Graph1.png
  
• Attachment: Graph2.png
  
• Attachment: zhi-h5-z25-x500-rou650-mw162-hc16744-0g-1.5s-5cms-changshi.fds

[gforney]

I believe I have found the problem. We were double counting the convective heat 
transfer from the surface. For most large fire applications, this term is so small

that it went unnoticed. In your case, it was significant. I committed the fix, and

it will appear in the next release. Please verify then if the case works properly.

Original issue reported on code.google.com by mcgratta on 2010-05-19 19:25:30

[gforney]

Thanks for your help.
But now I have another question for long time. We know the gas conduct is the main

heat transfer in opposed flame spread process. For low gas velocity the gas phase 
radiation and solid surface radiation will become important. We can get HRR, 
CONV_LOSS, COND_LOSS and RAD_LOSS in hrr file. But how to connect with each other?

Now I want to research the role of gas phase radiation and solid surface radiation

in my case.
I would be very happy for your advice.

Original issue reported on code.google.com by rentanx@yahoo.cn on 2010-05-20 14:37:11

[gforney]

As I said above, there is a bug in the calculation of CONV_LOSS. The actual FDS 
calculation is fine. The problem is only in the output quantity CONV_LOSS. It 
appears to me that in your case CONV_LOSS is very small, and COND_LOSS is almost the

same as HRR. This means that the heat from the flame is conducted and radiated to 
surrounding surfaces. RAD_LOSS indicates how much of that heat is radiative.

Original issue reported on code.google.com by mcgratta on 2010-05-20 15:04:48

[gforney]

I am closing this case.

Original issue reported on code.google.com by mcgratta on 2010-07-22 22:01:43

[gforney]

You said it would appear HRR=CONV_LOSS+COND_LOSS in the next release. Now the version
of the FDS can do this now? Thank you very much.

Original issue reported on code.google.com by rentanx@yahoo.cn on 2010-07-23 04:56:14

[gforney]

Are you asking a question? The question mark (?) at the end of a sentence turns that
statement into a question, like the "ma" at the end of a Chinese sentence. Are you
saying that you have verified that FDS is working properly with the latest release,
or are you asking if it works properly with the latest release?

Original issue reported on code.google.com by mcgratta on 2010-07-23 11:24:48

[gforney]

I am sorry. I am just asking if it works properly with the latest release? Can you give
a answer to me?

Original issue reported on code.google.com by rentanx@yahoo.cn on 2010-07-23 23:16:18

[gforney]

I would like you to test the case again. This is how we verify that the bug is fixed.

Original issue reported on code.google.com by mcgratta on 2010-07-26 11:57:32

[gforney]

Thanks for your advice. I will test the case as soon as possible.

Original issue reported on code.google.com by rentanx@yahoo.cn on 2010-07-27 04:45:42

[gforney]

I test the case again using FDS 5.5.1 amd now HRR=CONV_LOSS+COND_LOSS.
Thank you very much.

Original issue reported on code.google.com by rentanx@yahoo.cn on 2010-08-03 01:41:20

---

• Attachment: fds5.5cal.bmp

[gforney]

Thanks for checking this.

Original issue reported on code.google.com by mcgratta on 2010-08-03 12:02:03