Finite Element-based Stress Intensity Factor Estimation for Fatigue Crack Growth Simulation and Stochastic Filtering-based Fatigue Crack Growth Prediction of Pipelines

• Author / Creator

  Bartaula, Durlabh

• Fatigue cracking is one of the major integrity threats to oil and gas pipelines. Having a reliable trajectory of fatigue crack growth once detected is very crucial for decision making regarding integrity threat management. In-line Inspection (ILI) tools, and other non-destructive methods are used to assess different damage levels, such as measuring fatigue crack sizes in pipelines. Furthermore, fracture mechanics-based models, such as linear elastic fracture mechanics (LEFM) based Paris’ law, is used for prediction of future crack growth trajectory for fatigue crack integrity management in oil and gas pipelines.
  
  Neither fracture mechanics-based models nor crack measurements can be solely used to make a perfect fatigue crack growth (FCG) related integrity threats management decisions. However, the information contained in both sources can be fused to make better FCG predictions to support integrity management decisions. As such stochastic filtering specifically Particle Filter (PF), an iterative Bayesian approach, can be used for extraction of information about unknown model parameters like the material properties and crack sizes at certain time of interest by using data measured up to and including that point of time. Afterwards, Paris’ law can be used to predict future trajectory based on the updated information from the PF-based estimation process. As such, a methodology to couple the Particle Filter and Paris’ law, stochastic filtering-based FCG prediction, is developed in this study as a tool for pipelines with a fatigue crack.
  In the Paris law, the range of Stress Intensity Factor (SIF), the other important parameter besides the material fatigue crack resistance properties, are usually estimated using industry standard codes such as API 579 or BS 7910. In this study, the fatigue crack driving parameter SIF calculation using extended finite element method (XFEM), as well as conventional finite element method (FEM) implemented in Abaqus®, was dealt in detail along with various factors, like mesh size, mesh/element type, number of contour request and enrichment radius at the crack tip/front. The SIF estimation results were compared with the aforementioned industry models for cracked pipelines and analytical solutions for compact tension (CT) specimens. This aims to explore the capability of XFEM for SIF estimation in cracked pipelines, and also to assess the accuracy of the industry models (e.g., API 579). An indirect method of estimating fatigue crack growth (FCG) trajectory incorporating the SIF estimated using XFEM is then demonstrated for both a CT specimen and pipeline section tested in the literature. FCG trajectory, which is very sensitive to the fatigue parameters and the SIF, was successfully estimated reasonably well.
  Instead of using the API 579 model, directly for stochastic filtering-based FCG prediction process, a single surrogate model using Gaussian Process Regression (GPR) was developed and used. The GPR model, trained using validated SIF data points based on the various numerical and analytical models, was built for quick, reliable and computationally less expensive SIF estimation.
  To this end, a PF-based fatigue crack growth prediction methodology for pipelines was developed to leverage measurement data and known physics-based model, e.g., Paris law. This methodology was implemented as a python tool, to (1) jointly estimate fatigue model parameters and crack sizes, and (2) predict future trajectory of the fatigue crack growth in pipelines. This tool was applied to case studies using synthetic data of noisy crack size measurements in a pipeline for the purpose of demonstration.

• Subjects / Keywords

  • XFEM
  • FEM
  • PF
  • GPR
  • Pipeline
  • Integrity
  • SIF
  • ILI
  • Fatigue crack growth prediction
  • Measurement uncertainty

• Graduation date

  Spring 2022

• Type of Item

  Thesis

• Degree

  Master of Science

• DOI

  https://doi.org/10.7939/r3-gkpd-1j45

• License

  This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.