Numerical Instability

[GoogleCodeExporter]

Please complete the following lines...

Application Version:  5.1.6 Serial
SVN Revision Number: 1673
Compile Date:
Operating System: Linux

Describe details of the issue below:

Hello,

I am trying to simulate pool pan benzene fire with crib of wood. I have
encountered NUMERICAL STABILITY problem after 24 second. However, I modeled
the wood as has been already described with char_solid.fds case. The
capacity of the pool is 1 liter. 

Thanks in advance

Original issue reported on code.google.com by aymanhem...@googlemail.com on 2 Jul 2009 at 12:02

Attachments:

• wood.fds

[GoogleCodeExporter]

You cannot use parameters like EXTERNAL_FLUX and H_FIXED in a simulation like 
yours. 
Those parameters are used in the special case where we only treat the solid 
phase. 
Tell me if this solves the problem, but make sure you go through all the 
parameters 
to make sure you understand them. Do not just blindly copy input lines.

Original comment by mcgra...@gmail.com on 2 Jul 2009 at 12:22

• Changed state: MoreInfo

[GoogleCodeExporter]

Original comment by mcgra...@gmail.com on 2 Jul 2009 at 12:22

[GoogleCodeExporter]

I have already modeled pyrolysis including wood with thermal properties k, rho 
and c 
on MATL line. Then, MLRUA and IGNITION_TEMPERATURE on the SURF line. 
Unfortunately, 
I have encounterd the same problem after about 5 iterations. On the other hand, 
If I 
am working with burner as a fire source not a pool pan fire, it is working. I 
do not 
know if the problem with modelling wood or pool pan fire.

Original comment by aymanhem...@googlemail.com on 2 Jul 2009 at 4:32

Attachments:

• wood.fds

[GoogleCodeExporter]

Does the attached file fail after 5 iterations?

Also, have you tried using FDS 5.3.1?

Original comment by mcgra...@gmail.com on 2 Jul 2009 at 5:00

[GoogleCodeExporter]

You cannot include only MLPUA and IGNITION_TEMPERATURE on the SURF line -- 
there is 
no way to know how rapidly the object would heat up. You need a MATL_ID and a 
THICKNESS too. Tell me if that helps.

Original comment by mcgra...@gmail.com on 2 Jul 2009 at 5:11

[GoogleCodeExporter]

I am turning the issue over to Simo Hostikka.

Simo -- there are various problems with the file wood.fds, but I noticed in the 
latest version that YPRSUM can be zero and lead to an error where PPSURF=0 in 
the 
denominator of an expression.

Original comment by mcgra...@gmail.com on 2 Jul 2009 at 5:25

[GoogleCodeExporter]

The gasoline material did not have HEAT_OF_REACTION. Zero HoR affects the
Clasius-Clapeyron estimate of vapor pressure. I added a checking for this into 
read.f90.

Kevin -- I also fixed the potential divide-by-zero by YPRSUM.

The file runs now, at least up to 60 s. The updates will appear in FDS 5.4.0

Original comment by shost...@gmail.com on 3 Jul 2009 at 8:46

• Changed state: Fixed

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

I have added the HEAT_OF_REACTION. However, I have not encountered NUMERICAL 
INSTABILITY problem. I have another problem that the crip of wood did not burn. 
I 
noticed that the HRR increaed and decreased to zero after 60 second. 

Original comment by aymanhem...@googlemail.com on 3 Jul 2009 at 11:08

Attachments:

• wood.fds

[GoogleCodeExporter]

You haven't specified any product yields of your WOOD material.

Original comment by shost...@gmail.com on 3 Jul 2009 at 11:57

[GoogleCodeExporter]

I have already specified products yelds as shown in attached new file 
(wood.fds) 
with the same problem. You can check the HRR in the figure attached with 
increasing 
and decreasing to zero. That means there is no burning to the wood crips. I 
have 
tested this problem with the followng

1- SPRUSE surface as defined in charring_solid.fds example 
2- CHAIR material with described directly the wood material

Unfortunately, I have encountered the same problem but with changing the value 
of 
HRR but with the same trend.

Original comment by aymanhem...@googlemail.com on 6 Jul 2009 at 7:51

Attachments:

• wood.fds
• HRR.png

[GoogleCodeExporter]

Is the problem (a) the wood does not burn, or (b) there is a numerical 
instability?

Original comment by mcgra...@gmail.com on 7 Jul 2009 at 11:53

• Changed state: MoreInfo

[GoogleCodeExporter]

The problem now is the wood does not burn.

Original comment by aymanhem...@googlemail.com on 7 Jul 2009 at 11:59

[GoogleCodeExporter]

Furthermore, pool fire has been extinguished suddenly after about 30 second. 

Original comment by aymanhem...@googlemail.com on 7 Jul 2009 at 12:05

[GoogleCodeExporter]

I suggest you use positive value for gasoline heat of reaction. Positive means 
endothermic evaporation reaction.

I also suggest that you should use the method presented in User's guide for 
assessing your pyrolysis models piece by piece. Now you put together very 
complicated fuel packages. Finding errors becomes inevitably very difficult.
First make sure your pool burns as you want it to burn. Then make sure your 
wood 
surface burns realistically under external_flux = 50.

As the original problem (numerical instability) is not an issue anymore, we 
will 
close this issue.

Original comment by shost...@gmail.com on 7 Jul 2009 at 12:46

[GoogleCodeExporter]

Original comment by shost...@gmail.com on 15 Sep 2009 at 11:25

• Changed state: Verified