| Summary: | 碩士 === 國立臺灣科技大學 === 機械工程研究所 === 83 === The aerodynamic characteristics of high speed flying vehicles
are affected by many complex flow phenomena such as shock
waves, boundary layer separation, vortex shedding, turbulence,
etc. In recent years, the revolutionary progress of
computational mechanics and the rapid development of computer
hardware/ software provide an optimistic future for the
realistic numerical simulation of these flow phenomena. The
purpose of this study is to apply the recently developed h-
adaptive finite elemen method to analyze inviscid, high speed,
axisymmetric flow problems. The adaptive procedures make use of
mathematically justified error indicator to evaluate the
"goodness" of the numerical solution and they subsequently
optimize the structure of the grid to deliver answers which
satisfy user-specified levels of accuracy. The computational
grid is adapted in order to reduce the error using a minimal
number of degrees of freedom; this is achieved by employing
nested local refinements (unrefinements) in regions of large
(small) error. This type of grid-solution adaptive scheme
automatically resolve shocks to a level of accuracy specified
by the user and it employs minimal computational effort to
achieve this task. The technical bottlenecks for implementing h-
adaptive finite element method are : the treatment of
constrained nodes and the design of associated data
structures. In this study, a robust data structures are
designed for efficiently implementing grid refinement and
unrefinement. By modifying the local shape functions and the
associated physical degrees of freedom in an element, the
constrained nodes can be completely removed. To verify the
proposed method, a series of high speed flow problems are
solved and compared with either analytical solution or
experimental data. The test results demonstrate that the
accuracy of the numerical solution can be significantly
improved by using the h-adaptive finite element method.
|
|---|
