[chrisf90]

First of all: Hi everybody as this is my first post in this forum.
I'm working with Fluent for about 2 months so I'm defently a newbie but a lot of answered questions in this forum makes the "getting started" process much more efficient. Thank you for that.

So to my problem: I am trying to simulate air flow and heat transfer in a double skin facade (DSF). This is a facade where a second layer (made out of glaz) is placed up to 1 or even 2 metres in front of the conventional facade. The outer layer has openings for air in- and outlet. When solar radiation passes the outer single glazing it is absorbed by the shading device which is located in the facades cavity. This is way the temperature of the air increases in the cavity and natural convection occurs.

I have attached a sketch of my problem's geometry. The boundary conditions are:
B1 pressure inlet 303K
B2 outlet
B3 semitransparent exterior wall with solar heat flux 1000W/m²
B4 adiabatic
B5 mixed thermal B.C. 298K for external radiation and 7.7 W/m²K for convection

The question which I am facing is: Why is the solar radiation flux at boundary B3 bent (refracted)? See attachment2. I want the radiation radiate orthogonally to B3 until it hits a solid in the domain or a external boundary. I am worried about that because there a regions (at the top of double glazing) where no solar radiation is adjacent.

In Fluent Theory Guide (chapter 5.3.6.7.5.) it says that the wall should have zero thickness and a non-zero absorption coefficient. So I modeled it with A=0.01 [1/m] because I don't want to calm down the intensity through the boundary. The refractive index of wall B3 is set to 1 as i don't want reflection back from B3. In Figure 5.12 in Theory Guide it also says that the refractive indices of outside the domain and the adjacent solid or fluid zone after the external wall should not be equal. Has anybody an explanation why?

But ok, as it says in Users Guide (chapter 13.3.7.3.2.) that the refractive index of outside the domain is assumed to be 1 I modeled n of air with 1,01.
I also tried it with both fixed to 1 and the solution was the same.

By the way:
Are there better opportunities to model solar radiation in 2D problems? As solar ray tracing (including solar load model) is only available in 3D, nothing better came to my mind than modelling as I did.

I would appreciate any advice and have to say sorry for my long introduction to my problem.

Cheers Chris