The Largest Open Access Portal of Nondestructive Testing (NDT)

The possibility of cracks and other flaws occurring transverse to the direction of welding must not be taken lightly. The detection of such flaws is always included in the training of UT operators but operators rarely experience these flaws in practice, so their skills can atrophy. Experience has shown that transverse cracks can and do occur for a variety of reasons and their detection may not be straightforward. This paper discusses the use and limitations of conventional UT methods and how new methods can contribute to a more effective inspection. Both Manual Ultrasonic Testing (MUT) and Automated Ultrasonic Testing (AUT) are considered.

We present the first attempt to perform short glass fiber semantic segmentation from X-ray computed tomography (CT) volumetric datasets at medium (3.9 µm isotropic) and low (8.3 µm isotropic) resolution using deep learning architectures. We performed experiments on both synthetic and real CT scans and evaluated deep fully convolutional architectures with both 2D and 3D kernels. Our artificial neural networks show robustness at processing medium resolution scans and outperform existing methods in low resolution.

Extracting each part of an assembled object from a CT volume is important for analyzing the assembled parts, e.g., evaluating part deformation, measuring the clearance between different parts, and checking the relative positions of the parts. However, separating parts sharing narrow gaps is difficult because the changes in the CT values across narrow gaps are too small to detect using conventional methods, e.g., peak detection of the gradient norms of the CT values. In this study, a method used in detecting a narrow gap whose size is comparable to the voxel-size is proposed. The proposed method is based on peak detection of the Hessian eigenvalues of the CT values. Combining the proposed method with a conventional gradient-based method, we developed a system to segment an assembly composed parts made of the same material.

The ability to manufacture complex internal features is one of the distinct differentiators of Additive Manufacturing (AM) as compared to other manufacturing methods for metal components. This manufacturing process provide designers with new opportunities in the design, such as e.g. networks and curved and non round cooling channels. To fully take advantage of metal AM in industrial use, robust methods for the detection of potential internal defects is however needed. A method that holds the promise of being one of the few tools for non-destructive evaluation (NDE) of internal features and defects is X-ray computed tomography (CT). The applicability and limitations of CT, especially for defect determination in products with complex internal structures is however not fully understood. In this work, parts with different sizes of controlled internal defects in the form of slots of varying width, 0,1 – 0,4 mm was manufactured by AM, using Selective Laser melting (SLM). The parts were produced in both titanium and aluminium alloys and both with internal networks and as solid pieces. For both of the designed types of samples, containing the pre designed defects, the ability to detect the defects by industrial computed tomography (CT) was evaluated. The evaluation of defects using CT data can be done by a trained operator. For solid components this can be done with some assistance of analysis modes that are available in comersial software. For components with complex internal structures, the result is more operator dependant and more work is needed to develop methods for CT inspection that can enable automation of the inspection process and/or to assist a trained operator.

Absorption edge tomography is a method which exploits the sudden change of the attenuation coefficient, when the photon energy crosses the absorption edge of an element. The beamline BAMline at BESSY II, which is operated by the Federal Institute for Materials Research and Testing, can provide a monochromatized beam in a photon energy range from 5 keV up to 80 keV with a bandwidth of 2%. Together with the microtomography setup, this enables differential tomography sensitive to any element with N >= 24 (Cr) by using an appropriate K- or L-edge in this range. Here, absorption edge tomography at the Yttrium edge is employed to perform a non-destructive 3D characterization of the microstructure of a high strength Mg-Y-Zn alloy. The long period stacking ordered (LPSO) phase which forms fibres in this material was extracted based on the Yttrium content and the fibre length distribution was analysed.