search

firemodels / fds

Fire Dynamics Simulator
https://pages.nist.gov/fds-smv/
Other
664 stars 624 forks source link

High temperatures of water drops #3434

Closed shostikk closed 7 years ago

shostikk commented 8 years ago

Simplified version of the case discussed in thread https://groups.google.com/forum/?utm_medium=email&utm_source=footer#!msg/fds-smv/DkiUmfI08AU/RCvQBM_uBAAJ

Some droplets are high temperature.

mist.fds.txt

drjfloyd commented 8 years ago

I have been playing with this a little as well. It appears to be an overshoot issue from the very small drop size. Slightly larger drops and everything seems to behave fine. The issue looks like it is related to the assumption that TMP_G is fixed when we compute TMP_DROP_NEW allowing TMP_DROP_NEW to overshoot the gas temperature. Probably need to dust off (again) the equilibrium evap method Randy wrote some notes on.

rmcdermo commented 8 years ago

Jason,

First step would be to get T_EQ and Y_EQ as gas phase output quantities (thought you had worked on that, but do not see it). Then we could see if GET_EQUILIBRIUM is behaving.

On Wed, Feb 3, 2016 at 3:09 PM, Jason Floyd notifications@github.com wrote:

I have been playing with this a little as well. It appears to be an overshoot issue from the very small drop size. Slightly larger drops and everything seems to behave fine. The issue looks like it is related to the assumption that TMP_G is fixed when we compute TMP_DROP_NEW allowing TMP_DROP_NEW to overshoot the gas temperature. Probably need to dust off (again) the equilibrium evap method Randy wrote some notes on.

— Reply to this email directly or view it on GitHub https://github.com/firemodels/fds-smv/issues/3434#issuecomment-179433310 .

mcgratta commented 8 years ago

Don't we let the temperature change as we process droplets? Search on TMP_INTERIM in part.f90. This variable is set to TMP (the gas temperature from the EoS) at the start of the processing loop, and it is updated with each droplet. Before I made some changes to how we handle the divergence term, we just used TMP to hold the interim gas temperature.

drjfloyd commented 8 years ago

Randy, We did have it at one point but it was stripped at your or Kevin's request in one of the recent releases in a purge of code not actively being used. I will need to add it back.

Kevin,

If the drop subtimesteps we update the gas temperature. We don't check to see how far the drop has overshot the current gas temp. In this case in the first time step after injection the drop cools by evaporation and becomes very small. In the next time step as it is now cooler than the gas, it is heating up and over shoots the gas temp. If I add a line to prohibit this (a hack just for testing) the problem goes away. On Feb 3, 2016 3:19 PM, "Randy McDermott" notifications@github.com wrote:

Jason,

First step would be to get T_EQ and Y_EQ as gas phase output quantities (thought you had worked on that, but do not see it). Then we could see if GET_EQUILIBRIUM is behaving.

On Wed, Feb 3, 2016 at 3:09 PM, Jason Floyd notifications@github.com wrote:

I have been playing with this a little as well. It appears to be an overshoot issue from the very small drop size. Slightly larger drops and everything seems to behave fine. The issue looks like it is related to the assumption that TMP_G is fixed when we compute TMP_DROP_NEW allowing TMP_DROP_NEW to overshoot the gas temperature. Probably need to dust off (again) the equilibrium evap method Randy wrote some notes on.

— Reply to this email directly or view it on GitHub < https://github.com/firemodels/fds-smv/issues/3434#issuecomment-179433310> .

— Reply to this email directly or view it on GitHub https://github.com/firemodels/fds-smv/issues/3434#issuecomment-179439425 .

mcgratta commented 8 years ago

This commit adds a new wrinkle to this issue. Let's talk about it when you have a chance.

drjfloyd commented 8 years ago

I have pushed up the equilibrium routine discussed above under the flag EQUILIBRIUM_MODEL on mist (note at the moment the routine cannot handle drops on walls). The new evaporation routine avoids the problem in Simo's example (the case injects 10 C mist into 30 C air).

Below are two images with and without the routine. Without we see temperatures increasing over the background temperature of 30 C. With the new routine, we only see temperatures that are colder .

mist_a_0010 mist_0010

Thibauddu commented 8 years ago

Hello everyone, Following drjfloyd suggestion, here are two examples of simple simulation models with fuel spray ignition that return "Numerical Instability" for 7_2 and "Numerical instability for Particle Energy transport" for 7_3. It is about two concentric solids with inlet/outlet, and a fuel spray projecting small (50 microns) N-DECANE droplets with high speed towards the inlet.

I put comments in 7_2, so it should be the easiest one to begin with.

7_2.txt 7_3.txt

Thanks a lot for your work! Thibaud

rmcdermo commented 7 years ago

Jason, Have you tried these test cases with latest code?

drjfloyd commented 7 years ago

I am pretty sure I did. I can run them again.

mcgratta commented 7 years ago

Simo -- can you check to see if the original issue is resolved?

shostikk commented 7 years ago

Yes, the original issue is solved. I cannot find files for cases 7.2 and 7.3.

Current Date : September 8, 2017 13:54:00 Version : FDS 6.5.3 Revision : FDS6.5.3-2398-g1aa88f9-dirty Revision Date : Sun Sep 3 13:05:09 2017 -0400 Compilation Date : to 07.09.2017 08:23