Friday, May 18, 2012

Improved stl-file

I have made the stl 'watertight'. I make my stl-files in Rhino. Rhino gives this tip on stl:

1 Join the mesh objects.
Conceptually, this command gets all the triangles into one bag, but it doesn't glue the edges together. (The situation is similar to having surfaces that all fit together but have not been joined into a solid.) 
2) Weld the new mesh object.
3) At the Angle tolerance prompt type 180.
An angle tolerance of 180 tells the Weld command to glue adjacent triangle points together no matter what. 
4) UnifyMeshNormals.
This changes all the triangles so they are oriented the same way, that is, if two triangles share an edge, then they have the same idea of up.

It could be that using a watertight stl improves the mesh quality but I found some holes at the bottom... (You have to click on the picture to enlarge or look at the slice) Rhino reported it changed the normals of a quite lot of cells...

Looks like a perfect mesh: no bumps or holes
The trailing edge is pretty sharp

Oops holes !
A slice through the bumpy zone

Experiment with snappyHexMesh

I have created a simple knotted sphere in Rhino. Snappy generates a perfect mesh now... I used the same snappyHexMeshDict as in previous meshing of my guppy.

edge refinement at the cut

The red line is the edge for the Explicit feature edge refinement
Later I have done layer addition: 10 layers absolute thickness, expansion ratio 1.2. The result is not perfect to my lay eye.

The layer addition is not perfect

Thursday, May 17, 2012

meshing with featureEdgeMesh

With the snappyHexMesh utility a mesh can be constructed from a stl-file. I have extracted edges from the same stl-file with the surfaceFeatureExtract utility. These edges are used in the 'Explicit feature edge refinement' phase of the snappyHexMesh utility. This improves the quality of the edges of the mesh. But how can it be that the mesh is smaller than the edges extracted from the original stl-file ?

Another problem are the irregularities on the body. What went wrong? Is it the quality of my stl-file?

The mesh is smaller than the original stl-file !

 Two irregularities are visible.

a slice through the bumps

Original file (iges exported from proE)
holes on the top...
Yes they really are there
no mesh leaked inside

Tuesday, May 15, 2012

Thickness of the boundary layer

See this wiki page

For laminar boundary layers over a flat plate, the Blasius solution gives:
 \delta \approx 4.91 \sqrt{ {\nu x}\over u_0}
 \delta \approx 4.91x/ \sqrt{R e_x}
For turbulent boundary layers over a flat plate, the boundary layer thickness is given by:
 \delta \approx 0.382x/ {R e_x}^{1/5}
 R e_x = \rho u_0 x/\mu
 the overall thickness (or height) of the boundary layer
\nu is the kinematic viscosity
x is the distance downstream from the start of the boundary layer
Re_x is the Reynolds Number
\rho is the density
u_0 is the freestream velocity
\mu is the dynamic viscosit
My parameters
u_0~10m/s, \nu~1.5e-6, x~2 m.

 R e_x = \rho u_0 x/\mu
with \rho/\mu = \nu
Re_x=u_0 * x / \nu=1.33e7

\deltalaminar=0.0027 m = 2.7 mm
\deltaturbulent=0.0287 m =28.7 mm

In my mesh I will make 30 layers on the body with a total thickness of 30 mm. The thinnest layer will be 0.3 mm thick. With an expansionratio of 1.15 I will need 33 layers for 30 mm