Surface crack shape evolution modelling using an RMS SIF approach

A. Chahardehi, F. P. Brennan, S. K. Han

Research output: Contribution to journalArticlepeer-review

9 Citations (Scopus)


The majority of fatigue cracks in thick plate and tubular sections in structural components are two-dimensional surface cracks having significant propagation lives before becoming critical. The modelling of surface crack propagation life is important across a range of industries from power generation to offshore so that inspection, maintenance and repair strategies can be developed. Linear elastic fracture mechanics based predictions are commonplace, however, unlike thin sections with associated one-dimensional cracks frequently encountered in aerospace industries, crack shape or aspect ratio has a profound effect on crack front stress intensity factor and any resulting Paris Law based life prediction. The two most commonly used approaches are to calculate the crack growth rate at a number of points around the crack front and to consider only surface and deepest points, calculating the relative crack growth rates. Experience using these approaches has shown that the Paris Law coefficient as determined from plane stress specimens appears to lead to previously unpredicted inaccuracies. While this may suggest that the Paris Law is not suitable for this type of cracks, it is believed that a modification in the Paris Law would alleviate this problem. This paper examines this apparent anomaly, explaining why this discrepancy exists and suggests a practical solution using an RMS SIF approach for surface cracks that negates the need to correct the plane stress Paris Law constants.

Original languageEnglish
Pages (from-to)297-301
Number of pages5
JournalInternational Journal of Fatigue
Issue number2
Publication statusPublished - 1 Feb 2010
Externally publishedYes


  • crack aspect ratio
  • crack shape
  • Paris Law
  • stress intensity factor
  • surface crack


Dive into the research topics of 'Surface crack shape evolution modelling using an RMS SIF approach'. Together they form a unique fingerprint.

Cite this