Application of $J$-Integral to a Random Elastic Medium

TL;DR

This work addresses computing the statistics of the energy release rate $\mathcal{G}$ in a random elastic medium by applying the $J$-integral. The authors model the spatial variability of the Young's modulus as a homogeneous lognormal field $E(x)$ and use Monte Carlo with a modified contour integral to estimate the first two moments of $\mathcal{G}$, comparing results with those from the complementary energy method. They show that when $E(x)$ is homogeneous, the mean $J_*$ equals the mean $J$, so the mean fracture response is captured by Rice's $J$-integral, while higher moments are path-dependent; the correlation length $\ell$ mainly affects the second moment, with smaller $\ell$ reducing the coefficient of variation. Numerical SENT simulations indicate the mean aligns with the Irwin estimate $K_I^2 \langle 1/E\rangle$ and that a domain-integral correction is needed to reliably predict variance for arbitrary contours, highlighting implications for reliability-based fracture analysis of heterogeneous materials.

Abstract

This study investigates the use of the $J$-integral to compute the statistics of the energy release rate of a random elastic medium. The spatial variability of the elastic modulus is modeled as a homogeneous lognormal random field. Within the framework of Monte Carlo simulation, a modified contour integral is applied to evaluate the first and second statistical moments of the energy release rate. These results are compared with the energy release rate calculated from the potential energy function. The comparison shows that, if the random field of elastic modulus is homogeneous in space, the path independence of the classical $J$-integral remains valid for calculating the mean energy release rate. However, this path independence does not extend to the higher order statistical moments. The simulation further reveals the effect of the correlation length of the spatially varying elastic modulus on the energy release rate of the specimen.

Figures (4)

• Figure 1: Calculation of the energy release rate from an arbitrary domain containing the crack tip.
• Figure 2: Demonstration of the path independence property of $J_*$.
• Figure 3: Geometry of the SENT specimen and the different contours used for evaluation of $J$ and $J_*$.
• Figure 4: Typical realizations of the random field of Young's modulus for different correlation lengths.