[TurnbullJ]

In Appendix E of the DNV code 'DNV-RP-C205 ' the drag coefficient for a cylinder and a elliptical section are given as 1 and 0.6 respectively (for ellipse ratio of 2:1). The Reynolds number is given as 10^5

Using Ansys CFX 14, 2D steady state model, SST turbulence model, y+~5 and a slow mesh size growth rate away from the ellipse, I am not able to obtain a drag coefficient of above 0.3. I have refined the mesh both upstream and downstream of the ellipse.

The ellipse is vertically aligned and assumed long in the vertical direction. This has led to a 2D assumption in the modelling.

Inlet is a velocity boundary, outlet is pressure. Top and Bottom are symmetry and the sides are free slip walls. The 2D domain extends ~5 widths upstream and to the sides of the ellipse, and ~15 diameters downstream. The model is 1 cell thick in the vertical direction.

For larger 'timesteps' I get a steady solution and as I reduce the 'time' I do detect periodic vortex shedding from the ellipse. However, the Cd prediction remains low.

I'm sure that I am missing something, but can't see what. Suggestions gratefully received.

[hmasenger]

a pic of domain would really help .i dont understand why you used symmetry at top and bottom ?!it is usual to use this kind of BC for sides!i cant imagine the fluid domain..

[mvoss]

If you see reattachment- try out the GammaTheta-Model.

[ghorrocks]

Are you sure the flow is 2D at this Re? If the flow has 3D vortex shedding your 2D assumption will be significantly in error.