At the Hochschule Karlsruhe, Germany, a NACA model 4415 was tested in a wind tunnel by students as part of a fluid mechanical lecture. The experiment was done for different approach angles of the NACA-model. The experiment was also simulated with CFD-tools such as NOGRID points. We outline the physical problem and results obtained by NOGRID points for an approach angle of 20°. 

Physical model

The wind tunnel is modeled by a sufficiently large cuboid flow domain with "open" faces: Except for the inflow face the other five faces are outflow faces with a zero Dirichlet boundary condition for the pressure and zero Neumann boundary condition for the velocity. Using a "closed" box with only a small outflow area, opposite to the inflow area, would require a much larger fluid domain as otherwise the dynamic pressure would considerably increase.

At the inflow face the air flows in at 35.7 m/s in a circular area with a diameter of 0.35 m, outside the area the velocity is set to zero. This reflects the blast nozzle of the wind tunnel sufficiently well.

The faces of the NACA-model carry a wall slip condition. Compared to a noslip condition this yields much better numerical results as it avoids very large gradients of the velocity especially at the front of the model and it introduces only a small error.

Figure 1: A cut through the cuboid flow domain with the NACA model, the inflow is at the right hand side

The preprocessing required by the meshless CFD software NOGRID points mainly comprises the input of the above physical model parameters and the choosing of a smoothing length distribution, which effectively determines the density of points for the computation. As we are interested in the pressure profile at the surface of the NACA model we increase the density around the model. The high pressure gradient at the front of the model and the thinness at the back require an even higher density (or equivalently a smaller smoothing length). The distribution is shown in figure 2.

Figure 2: The smoothing length (which determines the point density) in the flow domain

Comparison of experiment and simulation

In figure 3 and 4 path integration yields stream traces, which show the typical wake vortex generated by the positive and negative dynamic pressure below and above the NACA-model, respectively.

In order to verify the simulation quantitatively as well, the measured pressure profile along the surface of the NACA-model is compared to the simulation. The experimental data and the simulation data are presented in figure 5. They show a very good agreement of the experiment and the simulation. The small differences are probably due to small simplifications done in the model such as the rather small fluid domain, the omission of the suspension gear of the model and the simplified blast nozzle of the wind tunnel.

Figure 3: NACA-Model with stream traces

Figure 4: NACA-Model with stream traces seen from behind the model

,                                

Figure 5: Comparison of experiment and simulation