CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   Questions about radiation modelling (

Yingchun October 19, 2005 12:13

Questions about radiation modelling
Hello experts,

I am modelling cavity airflows with radiation effects using CFX5.7. Within the cavity I need to model the effects of aluminium Venetian blinds. The blinds are very thin, not possible to model the thickness and heat transfer within large domain. I've been reading literatures and knew that some others were modelling the blinds surface as 'adiabatic'. I am bit lost here since as an 'adiabatic' boundary condition shown in CFX manual:

q_w = 0 = q_rad + q_cond.

Does this mean that after you set 'adiabatic' boundary on certain surfaces, the possible radiation emitted from the surfaces will be totally ignored? In 'heat transfer' option of cfx-pre, if you selected 'adiabatic', you can still have thermal radiation options to select, e.g. emissivity, does this mean that the radiation is still taking effects?

I am hoping this way: "a thin surface within an airflow domain is set as 'adiabatic' with an 'emissivity', the thin surface will take the adjacent air temperature as it's surface temperature (radiation power) to add the radiation effects", but not sure if this is the case in the code.

Thank you for your help.

Regards Yingchun

p.s. I asked this question in the following link but no response yet, so to ask here again.

opaque October 19, 2005 13:11

Re: Questions about radiation modelling
Dear Yingchun

The documentation implies that Adiabatic means "zero net heat transfer". It does not say anything about the different heat flux rates.. Therefore, the summation of all the heat flux rates at that boundary must vanish.

If you have radiation active and your emissivity is greater than zero (otherwise is a mirror), your radiative heat flux is non-zero.

Radiative heat flux is the difference between the emission and the incoming radiative flux (a.k.a irradiation).

The temperature at an adiabatic boundary is not the adjacent fluid temperature, but the one that allows that the sum of convective heat flux (coef1 * [Twall - Tadjacent]) and the radiative heat flux (coef2 * [sigma*Twall^4 - irradiation]) satifies the heat flux boundary condition (zero for adiabatic)

I am sorry I cannot comment on your question of how to models the blinds.

Good luck, Opaque.

Yingchun October 20, 2005 04:34

Re: Questions about radiation modelling
Thank you, Opaque, your comments do make a lot of senses to me about the adiabatic boundary. For the blinds, the experimental testing rig is going to measure the temperature, i may need to fix a temperature on the surface. Although the temperature (at some specific locations) may not represent the temperature of every single slat, it may provide a rough estimation.

There are two properties I am not quite sure how to define them when I set up the boundary conditions for the case. The thermal radiation model of 'Monte Carlo' is used. The model needs to provide a 'number of histories' but I am not sure what exactly this parameter means " in the manual it says 'optional parameter to indicate the total number of histories to be tracked for the simulation' , how to determine it and does it have anything to do with CPU time?

The other is under 'wall' boundary, selecting the thermal radiation option as 'Opaque', not quite understand how to set the 'diffuse fraction' an example the manual provides is for black body with unitary emissivity, this value is meaningless. To my understanding that if emissivity is 1.0 for an 'Opaque' solid surface, no matter what value you put for 'diffuse fraction' it isn't taking any effect for radiation modelling. But how about the emissivity is over 0 and less than 1.0? And in reality I suppose it's very difficult to judge the 'diffuse reflected energy' and the specularly reflected energy. Will the value of 1.0 for 'diffuse fraction' be sufficient for a normal surface, for example, aluminium with some coating leading to an emissivity about 0.6?

Thank you for your help

Regards, yingchun

opaque October 20, 2005 09:53

Re: Questions about radiation modelling
Dear YingChun,

The number of histories is the "discretization parameter" for the Monte Carlo model. The larger the number the "more accurate" your radiation field; however, the more expensive to compute as well. Since the MC model is an statistical model, a way to measure the quality of the solution is by looking at the standard deviation reported by the solver.

The standard deviation tends to reduce proportional to the square root of the ratio of histories between runs.

With the wall radiative characteristics, for black walls (e=1) the diffuse fraction is irrelevant. That is, why they are called black walls. For non-black walls (reflective walls), the diffuse fraction is a property that only the manufacturer, or modelling engineer can estimate. For example, aluminium sheets may have certain emissivity|diffuse fraction pair at the factory. After shipping, some humidity, rust/oxide/dirt/coating the properties at the factory are not as useful (except for a ballpark number).. You are the best source to know what value to plug in, or you can look at some heat transfer textbooks for some reference values.

Good luck, Opaque..

Yingchun October 20, 2005 10:06

Re: Questions about radiation modelling
Hi Opaque,

Thank you for your answer, you are really helpful and active on the radiation questions. thanks a lot

Regards, yingchun

abuthalib November 23, 2005 02:53

Re: Questions about radiation modelling *NM*

All times are GMT -4. The time now is 10:01.