Probabilistic nonlocal theory for quasibrittle fracture initiation and size effect. I: Theory

Zdenek P Bazant*, Drahomír Novák

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

93 Scopus citations


The nonlocal generalization of Weibull theory previously developed for structures that are either notched or fail only after the formation of a large crack is extended to predict the probability of failure of unnotched structures that reach the maximum load before a large crack forms, as is typical of the test of modulus of rupture (flexural strength). The probability of material failure at a material point is assumed to be a power function (characterized by the Weibull modulus and scaling parameter) of the average stress in the neighborhood of that point, the size of which is the material characteristic length. This indirectly imposes a spatial correlation. The model describes the deterministic size effect, which is caused by stress redistribution due to strain softening in the boundary layer of cracking with the associated energy release. As a basic check of soundness, it is proposed that for quasibrittle structures much larger than the fracture process zone or the characteristic length of material, the probabilistic model of failure must asymptotically reduce the Weibull theory with the weakest link model. The present theory satisfies this condition, but the classical stochastic finite-element models do not, which renders the use of these models for calculating loads of very small failure probabilities dubious. Numerical applications and comparisons to test results are left for Part II.

Original languageEnglish (US)
Pages (from-to)166-174
Number of pages9
JournalJournal of Engineering Mechanics
Issue number2
StatePublished - Feb 1 2000

ASJC Scopus subject areas

  • Mechanics of Materials
  • Mechanical Engineering

Fingerprint Dive into the research topics of 'Probabilistic nonlocal theory for quasibrittle fracture initiation and size effect. I: Theory'. Together they form a unique fingerprint.

Cite this