Start Date

2016 12:00 AM

Description

A common problem in industrial NDT is the accurate characterization of defects identified in ultrasonic array images. A particular challenge isthe distinguishing between volumetric defects (such as single gas pores and inclusions) and planar crack-like flaws. Compared with volumetric defects, crack-like defects are generally more severe from a structural integrity prospective and this raises demands for distinguishing these two defect types in both pre- service and in-service ultrasonic phased array inspection. In recent years, the use of ultrasonic arrays for non-destructive evaluation (NDE) has been revolutionized by of full matrix capture (FMC) which records the time-domain signals associated with every possible transmitter- receiver element combination and allows images to be generate in post-processing using different imaging algorithms, e.g., Total Focusing Method (TFM). In TFM, by exploiting reflections off geometric features, mode conversions at interfaces, and using different paths for transmitted and received waves, images showing multiple different views of the same region can be generated. These images are termed as multi-view TFM images and each of them corresponds to a specific wave mode combination (termed as imaging mode). In this paper, the feasibility of using the multi-view TFM images to distinguish between wavelength-size volumetric and crack-like defects is investigated through the analysis of TFM images from various simulated and experimentally-measured FMC array data sets. The data is from welded samples containing defects at the same location in the weld. The defects have the same size but different orientation angles and shapes. The simulated TFM images are normalized to the experimentally-measured noise image from the same imaging mode to indicate the Signal-toNoise Ratio (SNR) of each defect in each imaging mode. The presence of a defect of a particular type will typically be observable in some or all of the views with different SNRs. The different SNRs is the interplay between array element directivity, beam spread, transmission and reflection from geometric interfaces, scattering from material microstructure and scattering from the defect. Different types of defects have different imaging modes which give large SNR and this can be used to classify the defect type. Finally, the use of this approach is demonstrated in the experimental inspection of welded pipe samples.

Language

en

File Format

application/pdf

Share

COinS
 
Jan 1st, 12:00 AM

Distinguishing between Wavelength-size Volumetric and Crack-like Defects Using Multi-view Total Focusing Method Images

A common problem in industrial NDT is the accurate characterization of defects identified in ultrasonic array images. A particular challenge isthe distinguishing between volumetric defects (such as single gas pores and inclusions) and planar crack-like flaws. Compared with volumetric defects, crack-like defects are generally more severe from a structural integrity prospective and this raises demands for distinguishing these two defect types in both pre- service and in-service ultrasonic phased array inspection. In recent years, the use of ultrasonic arrays for non-destructive evaluation (NDE) has been revolutionized by of full matrix capture (FMC) which records the time-domain signals associated with every possible transmitter- receiver element combination and allows images to be generate in post-processing using different imaging algorithms, e.g., Total Focusing Method (TFM). In TFM, by exploiting reflections off geometric features, mode conversions at interfaces, and using different paths for transmitted and received waves, images showing multiple different views of the same region can be generated. These images are termed as multi-view TFM images and each of them corresponds to a specific wave mode combination (termed as imaging mode). In this paper, the feasibility of using the multi-view TFM images to distinguish between wavelength-size volumetric and crack-like defects is investigated through the analysis of TFM images from various simulated and experimentally-measured FMC array data sets. The data is from welded samples containing defects at the same location in the weld. The defects have the same size but different orientation angles and shapes. The simulated TFM images are normalized to the experimentally-measured noise image from the same imaging mode to indicate the Signal-toNoise Ratio (SNR) of each defect in each imaging mode. The presence of a defect of a particular type will typically be observable in some or all of the views with different SNRs. The different SNRs is the interplay between array element directivity, beam spread, transmission and reflection from geometric interfaces, scattering from material microstructure and scattering from the defect. Different types of defects have different imaging modes which give large SNR and this can be used to classify the defect type. Finally, the use of this approach is demonstrated in the experimental inspection of welded pipe samples.