It has been recognized that future intelligent design support environm
ents need to reason about the geometry of products and to evaluate pro
duct functionality and performance against given constraints. A first
step towards this goal is to provide a more robust information model w
hich directly relates to design functionality or manufacturing charact
eristics, on which reasoning can be carried out. This has motivated re
search on feature-based modelling and reasoning. In this paper, an app
roach is presented to geometric reasoning based on graph grammar parsi
ng. Our work combines methodologies from both design by features and f
eature recognition. A graph grammar is used to represent and manipulat
e features and geometric constraints. Geometric constraints are used w
ithin symbolical definitions of features and also to define relative p
osition and orientation of features. The graph grammar parsing is inco
rporated with knowledge-based inference to derive feature information
and propagate constraints. This approach can be used for the transform
ation of feature information and to deal with feature interaction.