Damage detection in flexible risers using statistical pattern recognition techniques

Carlos Alberto Riveros, Tomoaki Utsunomiya, Katsuya Maeda, Kazuaki Itoh

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

A statistical pattern recognition technique based on time series analysis of vibration data is presented in this paper. A 20-meter riser model experimentally validated is used for the numerical implementation of this technique. The dynamic response of the riser model is assessed using a semi-empirical approach with an increased mean drag coefficient model during lock-in events. Structural damage is associated with fatigue damage. Therefore, hinge connections are used to represent several damage scenarios. Then, the statistical pattern recognition technique is used to identify and locate structural damage using vibration data collected from sensors strategically located. Sensor locations are obtained from an optimum sensor placement method. The numerical results show that damage in oscillating flexible risers can be assessed using the presented statistical pattern recognition technique.

Original languageEnglish
Title of host publicationProceedings of The Seventeenth 2007 International Offshore and Polar Engineering Conference, ISOPE 2007
Pages2746-2753
Number of pages8
Publication statusPublished - 2007
Externally publishedYes
Event17th 2007 International Offshore and Polar Engineering Conference, ISOPE 2007 - Lisbon, Portugal
Duration: Jul 1 2007Jul 6 2007

Publication series

NameProceedings of the International Offshore and Polar Engineering Conference
ISSN (Print)1098-6189
ISSN (Electronic)1555-1792

Other

Other17th 2007 International Offshore and Polar Engineering Conference, ISOPE 2007
CountryPortugal
CityLisbon
Period7/1/077/6/07

All Science Journal Classification (ASJC) codes

  • Energy Engineering and Power Technology
  • Ocean Engineering
  • Mechanical Engineering

Fingerprint Dive into the research topics of 'Damage detection in flexible risers using statistical pattern recognition techniques'. Together they form a unique fingerprint.

Cite this