Start Date

2016 12:00 AM

Description

Due to both significant primary loads (weight, pressure) and thermal transients, important supporting structures (bedspring, deck…) subjected to thermo-mechanical fatigues require that periodical inspections are carried out in nuclear power plant.


The bedspring and the deck are complex welded structures of very restricted access; the ability to reliably detect and locate defects like cracks is therefore a difficult challenge. Ultrasonic testing is a well-recognized non-invasive technic which exhibits high characterization performances in homogeneous media (steel). However, its capabilities are hampered when operating in heterogeneous and anisotropic austenitic welds because of deviation and splitting of the ultrasonic beam.


In order to rise to this important challenge, a model-based method is proposed, which takes into account a prior knowledge corresponding to the welding procedure specifications (TIG welding, sequence of passes…) that condition the austenitic grains orientation within the weld and thus the wave propagation. The topological imaging method implemented is a differential approach which, compares signals from the reference defect-free medium to the inspected medium. It relies on combinations of two computed ultrasonic fields, one forward and one adjoint. Under the Born approximation, the investigated method virtually and successfully back-focuses the ultrasound residue towards any of the perturbations (defects) in the numerical model. However, when multiple reflections occurs, the conventional topological energy method fails at localizing the defect. This paper presents a new approach for addressing multiple reflexions. Numerical simulations and experiments have been carried out to validate the practical relevance of this approach to detect, locate and characterize an emerging notch both in a steel slab (consideration of the multiple reflections) and in a weld.

Language

en

File Format

application/pdf

Share

COinS
 
Jan 1st, 12:00 AM

Crack Detection and Localization inWeld Structure Using the Topological Energy Method

Due to both significant primary loads (weight, pressure) and thermal transients, important supporting structures (bedspring, deck…) subjected to thermo-mechanical fatigues require that periodical inspections are carried out in nuclear power plant.


The bedspring and the deck are complex welded structures of very restricted access; the ability to reliably detect and locate defects like cracks is therefore a difficult challenge. Ultrasonic testing is a well-recognized non-invasive technic which exhibits high characterization performances in homogeneous media (steel). However, its capabilities are hampered when operating in heterogeneous and anisotropic austenitic welds because of deviation and splitting of the ultrasonic beam.


In order to rise to this important challenge, a model-based method is proposed, which takes into account a prior knowledge corresponding to the welding procedure specifications (TIG welding, sequence of passes…) that condition the austenitic grains orientation within the weld and thus the wave propagation. The topological imaging method implemented is a differential approach which, compares signals from the reference defect-free medium to the inspected medium. It relies on combinations of two computed ultrasonic fields, one forward and one adjoint. Under the Born approximation, the investigated method virtually and successfully back-focuses the ultrasound residue towards any of the perturbations (defects) in the numerical model. However, when multiple reflections occurs, the conventional topological energy method fails at localizing the defect. This paper presents a new approach for addressing multiple reflexions. Numerical simulations and experiments have been carried out to validate the practical relevance of this approach to detect, locate and characterize an emerging notch both in a steel slab (consideration of the multiple reflections) and in a weld.