Novel XFEM Variable Strain Damage Model for Predicting Fracture in Small-scale SENT and Full-scale Pipeline Fracture Tests

  • Author / Creator
    Lin, Meng
  • Of all the common integrity threats of steel pipelines, cracking is the most dangerous and potentially resulting in the immediate loss of pressure containment capacity. Industry pipeline operators should ensure that the tensile train capacity (TSC) of welded pipeline exceeds the longitudinal tensile strains caused by substantial bending and/or tension due to external loads such as the seasonal ground temperature difference and soil differential movement. The tearing resistance curve (R-curve), such as the J-integral against the crack extension, quantifying the material’s inherent resistance to fracture, plays a critical role in predicting the TSC of welded pipeline. The R-curves of a pipeline material have recently been recommended to be measured from small-scale single edge notched tension (SENT) specimens, which can produce a relatively lower level of crack tip constraint similar to that of full-scale circumferentially cracked pipe specimens subjected to tension or even if it is globally loaded in bending. The extended finite element method (XFEM) has been increasingly implemented to predict the TSC of welded pipeline as well as the R-curves of SENT tests to assist with the pipeline integrity assessment. This method provides a robust approach allowing discontinuities such as cracks to be freely laid within the finite elements by introducing special enrichment functions. It alleviates the requirement for remeshing during crack propagation which is generally challenging to implement and computationally expensive in the conventional finite element method (FEM). The XFEM-based cohesive segments approach available in commercial finite element software Abaqus has been employed in predicting the crack initiation and propagation of pipelines since the last decade, but current damage criteria have not been well calibrated. Available criteria assume a fixed critical stress or strain value as the damage initiation, such as the maximum principal stress (Maxps) and maximum principal strain (Maxpe) damage criterion. Once an initiation criterion is met, the material cohesive stiffness is degraded with a specific damage evolution law till eventual failure, which can be simply characterized by the critical fracture energy release rate (G_c). However, the fracture behaviour predicted may be inaccurate due to its simplicity by ignoring important factors such as crack-tip constraint, which has a profound effect on fracture resistance. This doctoral thesis developed a novel XFEM variable strain-based damage criterion by introducing a variable critical strain profile as a function of stress triaxiality and Lode angle parameters accounting for crack-tip constraint. The new criterion was derived from the strain-based modified Mohr-Coulomb (MMC) fracture criterion developed for uncracked specimens and was applied to the small-scale SENT and full-scale pipeline fracture tests on cracked specimens. This criterion was implemented using Fortran programmed in Abaqus user subroutine-UDMGINI, and calibrated through models based on experimental data. The TSC of circumferentially surfaced-cracked X52 pipe specimens in full-scale pressurized tests and the J-R curves of X100 pipe specimens in small-scale SENT tests were well predicted with novel XFEM damage criterion. An optimal set of damage parameters (c_1 = 0.1, c_2 = 1.9, c_3 = 0.9, c_4 = 1, and G_c =200 N/mm) was calibrated specific to X52 with a given strain hardening exponent (n = 0.119) through 8 full-scale simulations based on experimental data of force against crack mouth opening displacement (CMOD) and end plate rotation, tensile strains along the pipe length at failure and fracture surface appearances. An optimal set of damage parameters (c_1 = 0.03, c_2 = 1.9, c_3 = 0.9, c_4 = 1, and G_c = 100 N/mm) was calibrated specific to X100 with a given strain hardening exponent (n = 0.0923) through 4 SENT simulations based on experimental data of force against CMOD and J-R curve. The effect of crack tip simulation using notch or planar crack and the differences of numerical predictions adopting fixed Maxps, fixed Maxpe or novel variable strain damage initiation criterion were carefully investigated in both X52 full-scale and X100 SENT models. This work is the first to couple XFEM with strain-based modified Mohr-Coulomb fracture criterion developed from uncracked specimens into full-scale and small-scale models of cracked specimens and validated from experiments. This work attempts to unify the classical fracture mechanics assuming a pre-existing crack and damage mechanics of uncracked bodies to form a unified theory in predicting fracture.

  • Subjects / Keywords
  • Graduation date
    Fall 2021
  • Type of Item
  • Degree
    Doctor of Philosophy
  • DOI
  • 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.