CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   CFX (https://www.cfd-online.com/Forums/cfx/)

 Mate April 17, 2003 09:33

Hello. I have a problem with thermal radiatiom. Radiation heat flux betwen two flat paralel surface is allways the same. If a change distance betwen surace or a change geometry of surface it is always the same. I use P1 radiatiom model in CFX 5.5. What influance have geometry factor F on radiatin in P1 radiation model. For air I set absorbtion=0.0001 scattering=0 and refractiv=1

Thanks

 Jeff Moder April 18, 2003 13:26

If your optical thickness is very small,

absorption*distance_btw_plates << 1

(which may be the case, i cannot tell from the info you provided) then you have an essentially transparent media, so of course the heat flux along the plate is independent of the distance between plates. If you want to see an effect, try making optical thickness significant, such as

absorption*distance_btw_plates = 1

Make sure your units are correct such that optical thickness is a non-dimensional number (such as 1/cm * cm).

I am not sure what you meant by change of surface geometry, but the total radiative heat load on surface 1 from surface 2 (and vice versa) will be independent of surface geometry IF

a) surface temperature, emissivity, reflectivity are constant along each surface

b) surface 1 can "see" all of surface 2 (and vice versa), that is F_12 = 1 and F_21 = 1

c) the medium between the surfaces is transparent

 Mate April 22, 2003 03:30

If we have vacuum between two surface then heat flux is q12=A1*F12*sigma*(T1^4-T2^4) and view factor F12 is dependent of surface geometrij and distance between them. I have two paralel plates 1m * 1m and distance betwen them is 1 m so F12 is approximatly 0.2 (Incropia)and not 1.

I think that if i take for absorption of air is 0.00001 1/m then the air will be like vacuum and P1 radiation method will be like surface-to-surface method but if P1 method always use F12=1 then it is not possible.

 Jeff Moder April 22, 2003 10:13

You say you have two paralel plates 1m * 1m and distance between them is 1 m.

Are you running this as a 3D or 2D simulation ?

What boundary conditions are you applying at the "open ends" of the computational domain (ie, the xmin and xmax boundaries in the sketch below) ?

c y

c ^

c |

c |

c ------------------------

c

c

c

c

c ------------------------- --> x

If you run the above problem as 2D in x-y, then the simulation will be modeling the plates as infinite in the z-direction, in which case F_12 = F_21 = 1 (for the plates). If the boundary condition at xmin and xmax is a black body at T=0, then we are back to the total heat load on each plate being independent of distance between plates.

If you run the above problem as 3D (ie, finite extent in z-direction, which for you case would be ymax-ymin = 1m, zmax-zmin = 1m), then you should see a dependence on the distance between plates. The accuracy of q12 is likely to be poor when using the P1 method on an essentially transparent medium. This is because the P1 method approximates the directional variation in the radiative intensity with a limited set of smooth functions (ie, a truncated set of spherical harmonics). This approximation works well WITHIN a strongly absorbing medium where the directional variation tends to be smooth, but typically does a poor job near boundaries where the intensity usually has strong variation in direction (eg, just think of difference of intensity for direction normal to plate going into plate versus going out of plate, and trying to model that sharp difference with a few sine or cosine functions).

 Mate April 22, 2003 10:33