Strong influence of air humidity on the heat flux

[mst-parkhachev]

Good afternoon, colleagues!

We have two almost identical projects. There is a large object on fire, and a vertical platform 10 m away. We are interested in the value of the heat flux on this platform. We considered two situations when the humidity was 0% and 40%.
humidity_0.txt
humidity_40.txt

In the first case, the air composition was as follows
&SPEC ID='AIR' BACKGROUND=.True. SPEC_ID(1:3)='OXYGEN','NITROGEN','CARBON DIOXIDE' VOLUME_FRACTION(1:3)=0.208638049437029, 0.79096861006154, 0.000393340501431118/

In the second case
&SPEC ID='AIR' BACKGROUND=.True. SPEC_ID(1:4)='OXYGEN','NITROGEN','WATER VAPOR','CARBON DIOXIDE' VOLUME_FRACTION(1:4)=0.206659977922372, 0.783469534601562, 0.0094808761872295, 0.000389611288837626/

It turned out that the heat flux in our model is extremely dependent on the air humidity. Two fluxes differ by almost three times. (The total heat release in both situations was the same.)

I don't know what the effect of air humidity is on radiation absorption in real fires. Is it really that big? Maybe something in our model is configured incorrectly?

[drjfloyd]

The default PATH_LENGTH for RADIATION is 10 cm. Using this over long distances over estimates the radiative absorption of atmospheric CO2 and WATER VAPOR.

1. You could increase the PATH_LENGTH
2. You could zero out the atmospheric CO2 and WATER VAPOR. Having atmospheric CO2 in the default AIR means let's us match experimental data when no combustion products are present. Having H2O in the default AIR helps prevent too rapid of droplet evaporation when sprinklers operate. If neither of those are a consideration for you, setting them to zero has a negligible effect on the temperature and velocity field that develop.

[mst-parkhachev]

For some purposes we need to know the air humidity in different parts of the building during a fire. Therefore, the ability to set the initial humidity is desirable for us.
I tried setting the PATH_LENGTH to 10 meters. Heat fluxes at 40% and 0% humidity became very close to each other.
Can we be sure that the heat flux at closer distances to the fire source will also be correct enough?
We are interested in distances of about one-two meters from the fire source and more. Up to several tens of meters.
Would it be a good idea for us to set the PATH_LENGTH equal to the diameter of the computational domain?
Can we assume that by setting PATH_LENGTH equal to, for example, 100 meters, at 40% humidity at any distance we will get approximately the same heat flow as at 0% humidity?

Does PATH_LENGTH also affect the absorption of thermal radiation by reaction products?

[drjfloyd]

You can always get the post-fire conditions with post processing. If you know the PRODUCTS mass fraction and the H2O from fire only mass fraction, then you can add back in the humidty as it will be in constant ratio to the PRODUCTS based on the AIR defintion with the HUMIDITY present.

A short path length will overestimate attenuation outside the flame at longer distances. The reverse is also true, that a long path length will underestimate at shorter distances. Atmospheric attenuation at 50 % humidty at 10 m is around 10 % and at 100 m is around 25 % (see Fig 66.36 in 5th SFPE Handbook). If you were to set the path length to the longest distance, your error at the smaller distances may not be that significant in comparison to the overall predictive uncertainties of FDS.

The path length impacts the absorption coefficients globally.

[mst-parkhachev]

Thanks for the detailed explanations!
The influence of PATH_LENGTH on the absorption of radiation by water vapor in the air is now more clear.
However, I am now concerned about the choice of PATH_LENGTH value in terms of its impact on absorption by the reaction products. Can we choose a universal value that will work well both close to the fire and far away?
In the case of a fire in an open area, the products move upward and occupy an area with a small diameter. In a fire in a building, the products move more horizontally and can spread quite far from the source of the fire.
If we set PATH_LENGTH to a large value, such as 10m or 20m, is it suitable for the interaction of radiation with reaction products when the area with products has a small radius (fire in open area)? Conversely, is the default value of 0.1m suitable for the case when the products are widely distributed in the compartment with fire?
In open areas, the task may be to measure the heat flux away from the smoke boundary, and in closed spaces, near the smoke boundary.
Will a PATH_LENGTH value of about 10-20 m be universal enough for our purposes? Unfortunately, we cannot customize the PATH_LENGTH value for each project individually, since we do not know exactly at what distance from the fire source the simulation results are of interest to our user.
Of course, errors of about 10% or comparable with the overall predictive uncertainties FDS will suit us.

[drjfloyd]

I suggest you run some test cases.