[chirag1993]

I'm modeling Rosseland diffusion radiation model in participating media in combustion model.
Geometry-
Half cylinder(fully fluid)
B.C.-
Three planes out of the total four planes are losing heat through radiation and convection to the atmosphere.
One middle plane is symmetry so my whole geometry will become a cylinder.
Modelling-
I have given scattering and absorption coefficients in the material properties and solving a transient 3D model. It is solving energy equation (for viscosity I have given Laminar cause I have no other choice because if I use a solid material I can't able to give any radiative properties in the material properties menu). I have put the initial condition for temperature for some region near to one plane which is higher than ignition temperature so it generates the heat will transfer.

Problem-
When I solved without radiation, it works perfectly but with radiation model, I got a floating error and sometime Divergence detected in AMG solver: Temperature error.

For that what I tried-
1)I tried relaxation factor still got the error
2)I varied time step with very high to low value still not worked
3)Also checked with varying initial condition for temperature except ignition temperature still not worked
4)Used different thermal conductivity still not worked
5)Gave initial condition little bit far from one plane still not worked

Only two things work-
1)Instead of giving the initial condition, I'm giving heat/energy at one plane and when it reach at sufficient temperature to liberate heat I changed the B.C. for that plane.
2)If I use small absorption and scattering coefficients and after some iterations, I again change the radiative properties according to my calculation, it works properly.

But I don't want to do such a manipulation, so can anyone help me to remove the error without manipulation.
Give your best ideas, so I can try.