A mesh-independent method for planar three-dimensional crack growth in finite domains

A. Sheng, N. M. Ghoniem, T. Crosby, G. Po

Research output: Contribution to journalArticlepeer-review

3 Scopus citations


The discrete crack mechanics (DCM) method is a dislocation-based crack modeling technique where cracks are constructed using Volterra dislocation loops. The method allows for the natural introduction of displacement discontinuities, avoiding numerically expensive techniques. Mesh dependence in existing computational modeling of crack growth is eliminated by utilizing a superposition procedure. The elastic field of cracks in finite bodies is separated into two parts: the infinite-medium solution of discrete dislocations and an finite element method solution of a correction problem that satisfies external boundary conditions. In the DCM, a crack is represented by a dislocation array with a fixed outer loop determining the crack tip position encompassing additional concentric loops free to expand or contract. Solving for the equilibrium positions of the inner loops gives the crack shape and stress field. The equation of motion governing the crack tip is developed for quasi-static growth problems. Convergence and accuracy of the DCM method are verified with two- and three-dimensional problems with well-known solutions. Crack growth is simulated under load and displacement (rotation) control. In the latter case, a semicircular surface crack in a bent prismatic beam is shown to change shape as it propagates inward, stopping as the imposed rotation is accommodated.

Original languageEnglish (US)
Pages (from-to)38-62
Number of pages25
JournalInternational Journal for Numerical Methods in Engineering
Issue number1
StatePublished - Jan 6 2019
Externally publishedYes


  • dislocations
  • finite element methods
  • fracture
  • solids

ASJC Scopus subject areas

  • Numerical Analysis
  • Engineering(all)
  • Applied Mathematics


Dive into the research topics of 'A mesh-independent method for planar three-dimensional crack growth in finite domains'. Together they form a unique fingerprint.

Cite this