Is there an estimate of the size of the stress term (especially the laplacian), in terms of the Reynolds-number? Formulated otherwise: how can it be, that Laplace(u)/Re can have an impact on the flow field? Laplace(u) is probably O(Re), but how can I see that?

Sincerely yours,

Philipp

It's not uniformly O(1/R)everywhere in the domain! If you write Y=y/R^(1/2) in the equations and take the limit R tends to infinity keeping Y=O(1) you obtain the Prandlt Boundary layer equations.

If a solution to these equations exists (and it usually doesn't because of flow separation) then the viscous correction is O(1/R^(1/2)) which is much bigger than your speculated O(1/R).

Now when the boundary layer separates the lack of a solution to the boundary layer equations indicates that the viscous correction is much greater than the above O(R^(-1/2)). For weakly (marginal) separated flow Triple-deck theory actually shows that the viscous correction is O(R^(-3/8)).

Hope his helps,

Tom.

Yes, that helped, thanks.

I understand that the impact of the viscousity in the boundary layer is significant. My problem is the simulation of fires in car tunnels and I'm thinking whether I have to include the viscous terms or not. The Reynold's number is about 10^6, so the boundary layer is very small. Is the impact noticable in the rest of the tunnel?

At a Reynolds number of O(10^6) your calculatation must be using a turbulence model. In which case you can get away with ignoring the "viscous terms". The uncertainty in the turbulence closure and the fire (combustion) model will be far greater in magnitude than the Reynolds number effects,

Tom.

Thanks, I will probably do that!

Yours,

Philipp