In essence, fracture mechanics consist of the conventional boundary-va
lue problem formulation of continuum mechanics, along with a variety o
f fracture criteria that govern the advance of crack fronts. The assum
ption that the material response is governed by a local constitutive m
odel everywhere in the body, even at points that are arbitrarily close
to the crack front, generally leads to an unbounded stress field as t
he crack front is approached. Also, processes such as metal-cutting an
d penetration share with fracture the essential feature of new-free-su
rface formation, but do not fit easily within the conventional fractur
e theory. For these and other reasons, it seems worthwhile to seek a b
roader theoretical construct which encompasses surface separation in m
ore general setting. With this motivation, a new theory is proposed wh
ich applies generally to surface separation in solid continua, but whi
ch nonetheless yields fracture-mechanics-type predictions in appropria
te special cases. The proposed exclusion region theory involves identi
fication of a small material neighborhood that contains the separation
front. A generalized constitutive description that derives directly f
rom the local constitutive model is constructed for the exclusion regi
on. A separation criterion is formulated with reference to tractions o
n, and/or distortion of, the exclusion region. The direction-of-advanc
e of the separation front is determined as a natural consequence of th
e separation criterion. The material parameters appearing in the separ
ation criterion can generally be determined from conventional fracture
tests. The theory has been implemented in a finite element code. Two
example problems illustrating certain important aspects of the theory
are presented. (C) 1997 Elsevier Science Ltd.