For structured grid generation, emphasis is placed upon creation of
grids with smooth, regular parametric and physical point distributions
that do not need to reflect the underlying topology and
parameterisation of the CAD model. This is desirable, since the CAD
model is often composed of a patchwork of ill-fitting surfaces, each
with their own mutually incompatible parameterisations.
Independence between the grid and the CAD surface parameterisation is
established at an early stage. Various techniques are available.
Central is the ability to evaluate a series of high quality sections
through the CAD model and to distribute points along them. Cases where
there are gaps between underlying surfaces are handled without difficulty.
Fine control over the density and distribution of points in both
parametric directions is usually a requirement in order to produce
the best results from the solver. The distribution may well be
driven by the surface curvature, but it also needs to depend on
the solver and the characteristics of the variable being solved.
Hence the best approach is to empower the GEMS user and provide
him with full and fine control over the placement of grid points.
A huge variety of point distribution methods are provided, based
upon physical arc length. These include cosine, cubic, exponential,
power law (all with various bunchings, scales and cut-offs) and local
curvature based distributions.
Of central interest to aerodynamicists is analysis of wings and other
shapes having aerofoil profiles, usually with smooth leading
edges. Grid generation around aerofoil sections is such a common and
important requirement that customised functions are available in GEMS
to ensure highest quality results. It is often desirable to bunch the
grid points symmetrically about the leading edge - but where is the
leading edge on a cambered profile? GEMS provides several methods
to evaluate leading edge position and subsequently to distribute grid
points about it. "Create Wing Grid" handles spanwise, and chordwise
distributions about the leading edge, in a single function.
|