[seb62]

Hi everyone,

I am running some 2D calculations over a flat plate, and comparing the openFOAM results with flu3m (Navier-Stokes CFD code from ONERA), Fluent and experimental/theoretical data. I am using 2 meshes.

Here is a description of the case :

Application : simpleFoam
Kinematic viscosity nu=1.57e-5 m²/s
Plate length L=1m
Re (end of plate) = 4.4e6
divScheme : Gauss upwind
Turbulence model : Spalart Allmaras
First mesh : 0.15 < Y+ (first node) < 0.2
Second mesh (identical to first mesh with 35 first cells close to the plate removed) :
20 < Y+ (first node) < 23

Boundary conditions :

Inlet : inlet
Top : pressureOutlet
Exit : extrapolatedOutlet
Plate : wallFunctions


U(inlet)=69.177m/s
nut(inlet)=1.5e-5 m²/s
nuTilda(inlet)=1.5e-5 m²/s
p(inlet)=p(top)=pref=0

Initial conditions = boundary conditions at inlet


For the first mesh :

Plots of U+=f(y+) :

At x=0.12 :




At x=0.94 :




For the second mesh :

Plots of U+=f(y+) and mut/mu=f(y+) :

At x=0.12 :



At x=0.94 :



Comments :

The mesh with y+ around 20 gives satisfying results especially on the velocity profile at x=0.94 and the local friction Cf. At x=0.12, we have an offset on the U+ velocity profile in the log region and a smaller turbulent viscosity compared to flu3m and Fluent

The finer mesh (y+ around 0.15) induces a significant offset on the U+ profile in the log region, with a smaller turbulent viscosity than the other calculations and an underestimated local friction (2.5e-3 compared to 3e-3 for the theoretical/experimental results at the end of the plate).

If someone encountered a similar problem...

Thanks in advance

[ivan_cozza]

Hi Sebastien,
I'm working on a similar case, a RANS over a flat plate at Mach 0.06 and Re 1.3x10^5.
If you choose a wall-function treatment, you have to put your first cell at y+ around 30, if you want to use a y+ less than 1, you have to use a lowRe model, so if find some lowRe model suitable for a flat plate you can use it!
Ivan