Dear All,

I am having difficulty implementing an isothermal wall boundary condition. My code is cell-centred, uses Roe's flux-difference approach, and the boundary conditions are imposed by a "ghost cell" method.

I have no problems with the adiabatic wall condition, which is imposed by ensuring that momentum components (rho*u and rho*v) in the ghost cell are opposite in sign to those in the neighbouring internal cell, while the density and the energy remain the same as in the internal cell. This ensures that there is no mass or energy flux through the wall

However I am unsure as to how to implement an isothermal wall using a similar approach. Has anyone any suggestions?

Kind Regards

Neil

Large gradient of T may appear on the wall for isothermal condition. You may try to set T(0,i.e.ghost)= 2*T_wall- T(1), with T_wall known.

While doing something along the lines of what Neil posted may work well, you may not, depending on the mesh at the wall get the correct gradients close to the wall.

I find that while this approach works ok for inviscid fluxe, for viscous fluxes, it is better to use the actual wall temperature value in calculating fluxes and gradients, rather than using the the internal face algorithm blindly. This will affect things like viscosity, heat flux calculation in the viscous fluxes.

I have seen that this gives a much better prediction of the wall heat flux value.

Hope this helps Srini

[SangeethCFD]

I found this paper which broadly discusses the same issue. In particular Section B of page 6 discusses this issue.

http://enu.kz/repository/2011/AIAA-2011-3381.pdf