From One-Month CFD To One-Day CFD - Efforts For Reducing Time and Cost of CFD

From One-Month CFD to One-Day CFD –
Efforts for Reducing Time and Cost of CFD –
K. Nakahashi, T. Fujita and Y. Ito
Department of Aerospace Engineering, Tohoku University,

Sendai 980-8579-01, Japan
Summary
This paper discusses our effort to develop a versatile and efficient CFD sys-
tem, named TAS-code (Tohoku University Aerodynamic Simulation Code).
The TAS code consists of several modules; surface mesh generation, tetrahe-
dral volume mesh generation, hybrid mesh generation, and flow solver. Stere-
olithography (STL) data are adopted as an interface between a CAD system
and the surface grid generator. The surface triangulation method uses the
advancing-front method coupled with geometric feature extraction technique
to enhance the capability to treat complex geometries. This approach signifi-
cantly reduces the work load to treat complex geometries. The flow solver for
Euler/Navier-Stokes equations is parallelized by the domain-decomposition
approach with MPI (Message Passing Interface) library. The mesh is par-
titioned by METIS with modification to create mesh overlapping between
adjacent sub-domains for information exchange. Load balancing for the hy-
brid unstructured mesh comprised of tetrahedrons, prisms and pyramids is
achieved by taking account of the number of edges in each sub-domain. The
capability of the TAS code is demonstrated for flow simulations around an
airplane and an insect.
1 Introduction
The computational fluid dynamics (CFD) has achieved a significant progress
in its algorithms and the applications in the last 30 years. In these days the
CFD is considered to be very close to its maturing stage. It is certainly true
that the CFD is routinely applied to analysis and design of airfoils/wings
of airplanes. For geometrically complex configurations, however, the current
CFD still needs painful efforts for the grid generation and the enormous
CPU times to obtain the desired accuracy level. Difficulty in generating a
grid around a complex geometry hinders the practical use of the CFD in
engineering analysis and design.
To overcome this, unstructured grids have become popular for complex
geometry CFD because of the easiness of the tetrahedral volume grid gen-
eration. However, even with the unstructured-mesh CFD, a time-consuming
From One-Month CFD to One-Day CFD 69
procedure is still required for generating a proper mesh on the 3D surface de-
fined by a CAD.This pre-process of the grid generation becomes more critical
as the complexity of the geometry increases. Another drawback of the un-
structured mesh CFD is its inherent overheads in memory and CPU time as
compared with those in structured meshes. The demand for reducing the to-
tal time required for flow analysis is increasing especially for aerodynamic
shape optimizations.
In this paper, our efforts to develop a versatile and efficient CFD sys-
tem, named TAS-code (Tohoku University Aerodynamic Simulation Code)
are described. Among the efforts, the surface mesh generation and the paral-
lelization of the flow solver, both are the key items to reduce the overall time
required for flow simulations, are discussed.
The capability of the TAS-code is demonstrated for two flow problems in
this paper. One is the flow computation around an experimental supersonic
airplane with a rocket booster. The other is a flow computation around an
insect with the fully detailed configuration.
2 Mesh Generation
2.1 Mesh Generation Procedure
Figure 1 shows a procedure of the unstructured mesh generation for flow com-
putations. Starting with the CAD output, the triangular mesh on the body
surface is generated at first. Then the tetrahedral volume mesh is generated in
the computational domain. For viscous flow computations, a hybrid unstruc-
tured mesh comprised of tetrahedrons, prisms and pyramids is generated.
The prismatic cells are used to resolve the thin boundary layers developed
along the surface.
In the present approach, the tetrahedral volume grid is generated by a
Delaunay-type generation method [1]. The hybrid unstructured mesh gen-
eration starts with isotropic tetrahedral mesh to enhance the robustness of
the algorithm [2]. These volume and hybrid mesh generation procedures are
basically automatic, once the surface and the boundary face meshes are de-
fined. The most time-consuming part of the mesh generation procedure for
complex geometries is the procedure from CAD data to the surface meshing.
2.2 Surface Mesh Generation

The detailed procedure of the surface meshing is described in [3, 4] and here
we discuss its outline. The basic strategy is to use both automatic and GUI-
based interactive procedures effectively. The automatic meshing is preferable
to reduce the user works. At the same time, the controllability of the local
grid density and quality by users is also very important for surface meshing
because the surface mesh directly affect to the solution accuracy.

From One-Month CFD To One-Day CFD - Efforts For Reducing Time and Cost of CFD

Uploaded by

Copyright:

Available Formats

From One-Month CFD To One-Day CFD - Efforts For Reducing Time and Cost of CFD

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

From One-Month CFD To One-Day CFD - Efforts For Reducing Time and Cost of CFD

Uploaded by

Copyright:

Available Formats

From One-Month CFD to One-Day CFD –

Eﬀorts for Reducing Time and Cost of CFD –

K. Nakahashi, T. Fujita and Y. Ito

Department of Aerospace Engineering, Tohoku University,

2.2 Surface Mesh Generation

You might also like