1. Introduction
Even though experimental observations could reveal atomic scale events, in principle, analytical predictions of atomic movements fall short of expectation by a wide margin. Classical dislocation models have shown to be inadequate by large scale computational schemes such as embedded atoms and molecular dynamics. Lacking in particular is a connection between interatomic (10
−8 cm) processes and behavior on mesoscopic scale (10
−4 cm) [1]. Relating microstructure entities to macroscopic properties may represent too wide of a gap. A finer scale range may be needed to understand the underlying physics. Segmentation in terms of lineal dimensions of 10
−6–10
−5, 10
−5–10
−3 and 10
−3–10
−2 cm may be required. They are referred to, respectively, as the micro-, meso- and macro-scale. Even though the atomistic simulation approach has gained wide acceptance in recent times, continuum mechanics remains as a power tool for modeling material behavior. Validity of the discrete and continuum approach at the different length scales has been discussed in [2 and 3].Material microstructure inhomogeneities such as lattice configurations, phase topologies, grain sizes, etc. suggest an uneven distribution of stored energy per unit volume. The size of the unit volume could be selected arbitrarily such as micro-, meso- or macroscopic. When the localized energy concentration level overcomes the microstructure integrity, a change of microstructure morphology could take place. This can be accompanied by a corresponding redistribution of the energy in the system. A unique correspondence between the material microstructure and energy density function is thus assumed [4]. Effects of material structure can be reflected by continuum mechanics in the constitutive relations as in [5 and 6] for piezoelectric materials.In what follows, the energy density packed in a narrow region of prospective crack nucleation sites, the width of this region will be used as a characteristic length parameter for analyzing the behavior of moving cracks in materials at the atomic, micro-, meso- and macroscopic scale level. Nonlinearity is confined to a zone local to the crack tip. The degree of nonlinearity can be adjusted by using two parameters (σ
0,ℓ) or (τ
0,ℓ) where σ
0 and τ
0 are referred to, respectively, as the stresses of “restraint” owing to the normal and shear action over a local zone of length ℓ. The physical interpretation of σ
0 and τ
0 should be distinguished from the “cohesive stress” and “yield stress” initiated by Barenblatt and Dugdale although the mathematics may be similar. The former has been regarded as intrinsic to the material microstructure (or interatomic force) while the latter is triggered by macroscopic external loading. Strictly speaking, they are both affected by the material microstructure and loading. The difference is that their pre-dominance occurs at different scale levels. Henceforth, the term restrain stress will be adopted. For simplicity, the stresses σ
0 and τ
0 will be taken as constants over the segment ℓ and they apply to the meso-scale range as well.
2. Elastodynamic equations and moving coordinates
Navier’s equation of motion is given by
(1)in which
u and
f are displacement and body force vector, respectively. Let the body force equal to zero, and introduce dilatational displacement potential φ(
x,
y,
t) and the distortional displacement potential ψ(
x,
y,
t) such that
(2)u=φ+×ψThis yields two wave equations as
(3)where
2 is the Laplacian in
x and
y while dot represents time differentiation. The dilatational and shear wave speeds are denoted by
cd and
cs, respectively.For a system of coordinates moving with velocity
v in the
x-direction,
(4)ξ=
x−
vt, η=
ythe potential function φ(
x,
y,
t) and ψ(
x,
y,
t) can be simplified to
(5)φ=φ(ξ,η), ψ=ψ(ξ,η)Eq. (3) can thus be rewritten as
(6)in which
(7)In view of Eqs. (7), φ and ψ would depend on (ξ,η) as
(8)φ(ξ,η)=Re[
F(ζ
d)], ψ(ξ,η)=Im[
G(ζ
s)]The arguments ζ
j(
j=d,s) are complex:
(9)ζ
j=ξ+iα
jη for
j=d,sThe stress and displacement components in terms of φ and ψ are given as
(10)uy(ξ,η)=−Im[α
dF′(ζ
d)+
G′(ζ
s)]The stresses are
(11)σ
xy(ξ,η)=−μ Im[2α
dF″(ζ
d)+(1+α
s2)
G″(ζ
s)]σ
xx(ξ,η)=μ Re[(1−α
s2+2α
d2)
F″(ζ
d)+2α
sG″(ζ
s)]σ
yy(ξ,η)=−μ Re[(1+α
s2)
F″(ζ
d)+2α
sG″(ζ
s)]with μ being the shear modulus of elasticity.
3. Moving crack with restrain stress zone
The local stress zone is introduced to represent nonlinearity; it can be normal or shear depending on whether the crack is under Mode I or Mode II loading. For Mode I, a uniform stress σ
∞ is applied at infinity while τ
∞ is for Mode II. The corresponding stress in the local zone of length ℓ are σ
0 are τ
0. They are shown in Fig. 1 for Mode I and Fig. 2 for Mode II. Assumed are the conditions in the Yoffé crack model. What occurs as positive at the leading crack edge, the negative is assumed to prevail at the trailing edge.
相似文献