Recent work has shown that parametrically formulated surface grid gene
ration algorithms are severely limited by solution bifurcations of the
discretized equations. Such bifurcations become apparent if the given
surface has points of high curvature. The present study demonstrates
that discrete equations derived from an alternative physical variables
formulation do not result in similar bifurcations. However, the natur
al formulation using physical variables is not entirely successful on
surfaces of high curvature due to excessive truncation error in the so
lution grid. It is shown that both bifurcations and truncation error a
re overcome by formulating the grid generation equations in terms of p
rojections of the physical variables onto the surface tangent plane.