Another key feature of the CCM solver is the ability to use a mesh consisting of arbitrary polyhedra. Aside from allowing the solver to handle any and every mesh type, the use of polyhedra with 12 or more faces results in a considerable memory and computing time savings over a pure tetrahedral mesh (about half the memory and one fifth of computing time was needed on average to achieve the same accuracy with polyhedra as with tetrahedra in some benchmark cases conducted by Dr. Peric).
While it is possible to convert any mesh into arbitrary polyhedra, certain benefits can be acheived using a grid generator tailored to make arbitrary polyhedral meshes. CD adapco Group has developed such a pre-processor, which is embeded in a solid modeler based on the Parasolid geometry kernel. The process begins by importing or creating a solid model of the geometry. Next meshing parameters, such as mesh spacing and prism layer height, are defined on the solid model. Finally, the mesh is created fully automatically.
From the same GUI it is possible to define the flow model, such as physical properties and boundary conditions, and launch the flow solver. Postprocessing is also done within the same environment.
The new CCM solver is scheduled for release in early 2004. At that time STAR-CD users will have the exciting option of trying a next generation simulation software.
I wish to thank Dr. Peric for his time and allowing me to visit his office. This story is being posted a considerable time after my visit, but the delay is due solely to my own inefficiency.
A model of a cyclone separator in the CCM pre-processor.
The polyhedral mesh of the cyclone separator. The region of the mesh
in the red rectangle is shown enlarged in the lower portion of the image.
Pressure on the surface of the cyclone separator.
Velocity vectors on the mid-plane of the cyclone separator.