Dear Gurus, please help me to solve the following problem:

TARGET: to map the convective heat transfer coefficient (H.T.C., h_cell, [W/m2 K]), without using the energy equation (the "cold" approximation I mean) using a cell by cell formulation, along the frontier of a 3D very complex geometry (many characteristic lengths, confined domain, forced convection)

Velocity and pressure field are known everywhere (already computed with a RANS FVM code). Lots of velocity gradients are present inside the whole domain. The flow is incompressible with M << 0.3 and isothermal, Pr = 0.7.

Someone told me to use the flat plate approximation computing for each wall cell the local Nusselt relation like this:

Nu(x) = c * (Re(x)^n )* Pr^m

where c, n and m are the laminar/turbulent coefficient suggested by Schlichting ("B.L. Theory" pag. 298-299)

Re(x) = density * vel_cell * x / viscosity

and substituting in:

h(x) = k * Nu(x) / x

to obtain the H.T.C., h(x).

The main problem is:

> what is "x" for my 3D CFD problem cell by cell based?

> Is a correct approximation to choose the wall cell mean length as "x" supposing that this dimension is inside the boundary layer (x=Lmean_cell)?

> What is the best choose for vel_cell?

Any contribution would be really appreciated.

Tes

it is relative to your flow direction

since you want to addopte a flate plate approximation, you should use the main streamdirction

I am making tests on a solar collector and for my theoretical determination of the absorbed energie I have to know the heat transfer coefficient at the outside of the collector to know how much energie I lose due to convection. Could somebody give me an aproximation for this coefficient when there is a little wind and when there is a lot of wind?