Reliability of Defect Detection in Welded Structures

Abstract

An analysis has been made of the ability of several nondestructive inspection procedures to reliably detect surface fatigue cracks. It is shown that the ability of penetrant systems of aluminum and titanium are over 90% for cracks whose surface length is greater than 0.939mm (0.037 in.) in length and decreases for cracks less than this value. The probability of detection of these cracks, given that the crack is present has been determined at 90, 95, and 99% confidence factors. The conditional probabilities associated with an error call have also been developed. For this analysis four outcomes were defined: (1) The detection of a crack that is present; (2) The missing of a crack that is present; (3) The detection of a crack that does not exist (false call); and (4) The verification that a part does not contain a flaw. Using Bayes theorem, the conditional probabilities have been established in terms of a material quality factor, an error probability ratio and a success probability ratio. The results of two inspectors, one with a 6% error probability, and the other with a 36% error probability have been examined.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jan 01, 1975
Accession Number
ADA019660

Entities

People

  • B. W. Yee
  • F. M. Wells
  • J. K. Malpani
  • P. F. Packman

Organizations

  • Vanderbilt University

Tags

DTIC Thesaurus Topics

  • Alloys
  • Aluminum
  • Aluminum Alloys
  • Defect Detection
  • Demonstrations
  • Detection
  • Distribution Curves
  • Distribution Functions
  • Equations
  • Inspection
  • Materials
  • Plastic Explosives
  • Probability
  • Reliability
  • Sensitivity
  • Titanium
  • Verification

Fields of Study

  • Engineering

Readers

  • Mathematical Modeling and Probability Theory.
  • Metallurgy
  • Systems Analysis and Design