I am working with wall functions with k-epsilon model. i am studying the wall heat transfer. With wall function approach, it is said that the boundary y+ should be about 30. So i maintained that and got some result.Now for doing grid independence, i halved each cell except the wall-nearest cells.i got a slightly different result..So i again halved all the cells except the wall-nearest cells to study grid independence..Now it became that the wall cells are larger than the ones away from the wall..Is it ok..What is the effect of this.. Any help is welcome.

When You calculate heat transfer problems, You must resolve two boundary layers: The momentum boundary layer and the thermal conduction boundary layer. The thickness of these two layers is, in general, different. In fluent, You fit the grid resolution only to the momentum boundary layer, thus the turbulent heat transfer may differ when You change the grid resolution.

My doubt is whether the wall funciton approach is enough when we study heat transfer to the wall or is there a need to go to the two-layer zonal model (fluent 5.3). with regards,

You should check first the non-dimensional thickness of the momentum (y^*) and thermal (y_T^*) boundary layers. In case of y^* < y_T^*, fluent uses a linear law in order to compute the heat flux through the thermal boundary layer, independent of the near-wall resolution of Your grid. Otherwise, You can try the two-layer model. Also try the RNG-k-eps turbulence model with the refined description of heat transfer.

hi i am also from iit madras. i have done grid independence study. first of all r u using fluent. for grid independence, dont try to play with cells near the walls. just double the no of cells by changing the scale factor in meshing option. then solve for both grids. most of the time dont try with convergence level but go for many no of iterations.