Numerical Instability Error

[esmatsh...@gmail.com]

Hi Everybody,

This is my first project using Pyrosim and FDS. I have modeled a pool fire caused by flammable liquid (size of pool fire is 9*2.5*0.2) in a dangerous goods building using Pyrosim. The building has got one storey and is divided to the different sections by the fire rated internal walls. The two sliding fire doors will be closed 45 seconds after fire detected by one of heat detectors (activation at 74 degree C). In addition, the air intake grille and air exhaust grille will be closed by fusible link at 93 degree C detecting by the gas devices provided closed to them. It took almost 19 hours to run!!! and I have finally got numerical instability error at 119 seconds in FDS and Pyrosim both.

If you could help me to address this issue,I would be grateful. I have attached a copy of FDS and OUT file for your review.

Regards,

Esmat  

[Randy McDermott]

First, thank you for working with the latest code!  If there are hidden issues this is the only way we can find them.

If this case was working in a version prior to v6.4, it is possible that our increase of the default VN number (in an attempt to speed up the code) is at fault (but no guarantee).  Try setting VN_MAX=0.5 and VN_MIN=0.4 on MISC.  This change is discussed in the release notes:

https://github.com/firemodels/fds-smv/wiki/FDS-Release-Notes

Option B is to tighten your velocity tolerance at a mesh boundary.  See the user guide.

--
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/e60e2c55-0912-4c73-98cf-4f770b275cd4%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Kevin]

Before doing that, consider this -- if you are going to seal your enclosure, you need to add a ZONE line

&ZONE XB=... /

This tells FDS that a portion of the computational domain is sealed and FDS will assume a different background pressure in this space.

Also, your meshing strategy is not good. The meshes are not equally sized, thus, if you run the calculation in parallel, the speed will be bound by the largest mesh (in the sense of number of grid cells).

On Friday, April 15, 2016 at 6:21:33 AM UTC-4, Randy McDermott wrote:

First, thank you for working with the latest code!  If there are hidden issues this is the only way we can find them.

If this case was working in a version prior to v6.4, it is possible that our increase of the default VN number (in an attempt to speed up the code) is at fault (but no guarantee).  Try setting VN_MAX=0.5 and VN_MIN=0.4 on MISC.  This change is discussed in the release notes:

https://github.com/firemodels/fds-smv/wiki/FDS-Release-Notes

Option B is to tighten your velocity tolerance at a mesh boundary.  See the user guide.

On Fri, Apr 15, 2016 at 1:26 AM, <esmatsh...@gmail.com> wrote:

Hi Everybody,

This is my first project using Pyrosim and FDS. I have modeled a pool fire caused by flammable liquid (size of pool fire is 9*2.5*0.2) in a dangerous goods building using Pyrosim. The building has got one storey and is divided to the different sections by the fire rated internal walls. The two sliding fire doors will be closed 45 seconds after fire detected by one of heat detectors (activation at 74 degree C). In addition, the air intake grille and air exhaust grille will be closed by fusible link at 93 degree C detecting by the gas devices provided closed to them. It took almost 19 hours to run!!! and I have finally got numerical instability error at 119 seconds in FDS and Pyrosim both.

If you could help me to address this issue,I would be grateful. I have attached a copy of FDS and OUT file for your review.

Regards,

Esmat  

--
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+unsubscribe@googlegroups.com.

[esmatsh...@gmail.com]

H Kevin,

Great thanks for your help.

I modified the meshes' size to be equal and I added a Zone inside the building. No more numerical instability error. However, I still got maximum velocity error. I think I should follow Randy's instruction to solve this problem as well. I'll come back to both of you when I try it. One more thing, Smokeview shows the pressure of the Zone inside the building as a constant value (-2.75 pa), does it make sense for you??

On Friday, April 15, 2016 at 8:43:00 PM UTC+8, Kevin wrote:

Before doing that, consider this -- if you are going to seal your enclosure, you need to add a ZONE line

&ZONE XB=... /

This tells FDS that a portion of the computational domain is sealed and FDS will assume a different background pressure in this space.

Also, your meshing strategy is not good. The meshes are not equally sized, thus, if you run the calculation in parallel, the speed will be bound by the largest mesh (in the sense of number of grid cells).

On Friday, April 15, 2016 at 6:21:33 AM UTC-4, Randy McDermott wrote:

First, thank you for working with the latest code!  If there are hidden issues this is the only way we can find them.

If this case was working in a version prior to v6.4, it is possible that our increase of the default VN number (in an attempt to speed up the code) is at fault (but no guarantee).  Try setting VN_MAX=0.5 and VN_MIN=0.4 on MISC.  This change is discussed in the release notes:

https://github.com/firemodels/fds-smv/wiki/FDS-Release-Notes

Option B is to tighten your velocity tolerance at a mesh boundary.  See the user guide.

On Fri, Apr 15, 2016 at 1:26 AM, <esmatsh...@gmail.com> wrote:

Hi Everybody,

This is my first project using Pyrosim and FDS. I have modeled a pool fire caused by flammable liquid (size of pool fire is 9*2.5*0.2) in a dangerous goods building using Pyrosim. The building has got one storey and is divided to the different sections by the fire rated internal walls. The two sliding fire doors will be closed 45 seconds after fire detected by one of heat detectors (activation at 74 degree C). In addition, the air intake grille and air exhaust grille will be closed by fusible link at 93 degree C detecting by the gas devices provided closed to them. It took almost 19 hours to run!!! and I have finally got numerical instability error at 119 seconds in FDS and Pyrosim both.

If you could help me to address this issue,I would be grateful. I have attached a copy of FDS and OUT file for your review.

Regards,

Esmat  

--
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.

[esmatsh...@gmail.com]

Hi Randy,

Thanks for your kind assistance. 

Based on Kevin's guidance below, I changed the size of meshes and I defined a Zone inside the building. By doing this, I run my project without numerical instability error. But I still got maximum velocity error that I think it will be solved if I follow your instruction. I'll come back to you while I run the project again.

[Kevin]

There is always a "maximum velocity error" associated with iterating the pressure solution mesh to mesh. You can tighten that tolerance if you want with VELOCITY_TOLERANCE on the PRES line.

[Andreas]

On Friday, April 15, 2016 at 12:21:33 PM UTC+2, Randy McDermott wrote:

First, thank you for working with the latest code!  If there are hidden issues this is the only way we can find them.

If this case was working in a version prior to v6.4, it is possible that our increase of the default VN number (in an attempt to speed up the code) is at fault (but no guarantee).  Try setting VN_MAX=0.5 and VN_MIN=0.4 on MISC. 

Hello,

I'm encountering stability problems since upgrading to V6.4.0 from V6.1.2. When running a model like the appended fds-file, the development of the HRR causes FDS to stop after a few seconds simulation time. With V6.1.2, the same model was simulated without any problems. I already tried to add VN_MIN=0.4 and VN_MAX=0.5 to the MISC-line like suggested above, but that doesn't seem to help.

... any suggestions on how to deal with this?

Many thanks in advance
Andreas

[Kevin]

Over the past week I have developed a way to reduce these numerical instabilities in tunnels. The problem is related to the decomposition of the pressure term in the momentum equation, the so-called baroclinic torque term. If all goes well, we will release a new version this week or next.