The transfinite interpolation formula of grid generation that employs
the Hermite interpolation formula is improved by the addition of a gri
d front marching method embedded with a simple grid smoother. In the t
est cases, the proposed method preserved approximate grid orthogonalit
y around all the boundaries and eliminates grid nonsmoothness, grid sl
ope discontinuity, and grid overlapping that occurred with the origina
l method. The required CPU time is about 2.2 times more than the CPU t
ime required by the original method and is at least one order of magni
tude faster than the elliptic grid solver employed as a grid smoother.
The method was also applied to the linear grid system to improve grid
smoothness, but its computing time is about 4 times greater than the
original method and is of the same order as the elliptic grid smoother
. For cases with boundary discontinuity, the grid front marching metho
d is applied to only a few grid levels inward, the remaining grids are
generated via the original transfinite interpolation formula, and the
required computing time is only slightly longer than that of the orig
inal formula.