Abstracts
8-10 September 2026
The International Centre, Telford, UK
[1A2] Assessing models for resin front position estimation using a network of ultrasonic guided wave sensors
C A Calistru¹, V Tunukovic¹, E Mohseni¹, S G Pierce¹, C N MacLeod¹, D Lines¹, R K W Vithanage¹, I Bomphray², T Weis², G Munro³ and T O’Hare⁴
¹University of Strathclyde, UK
²National Manufacturing Institute Scotland (NMIS), UK
³Boeing Aerospace Innovation Centre (AIC), UK
⁴Short Brothers, a Boeing Company, UK
Implementing out-of-autoclave (OoA) composite manufacturing at industrial scales requires reliable in-situ monitoring of resin flow. This study investigates the use of ultrasonic guided waves (UGWs) for real-time localisation of the resin front during liquid composite moulding processes using a non-intrusive sensing array. Through theoretical, simulation-based and experimental observations, the optimal strategy for monitoring liquid propagation through UGWs is identified. A custom liquid-only infusion mould allows repeatable trials and direct correlation between UGW signals and resin-front position. A refined finite element simulation workflow is developed to reproduce the fundamental modal behaviour of experimental waveforms at substantially reduced computational cost. A one-dimensional convolutional neural network trained directly on time-domain signals achieves a mean absolute error of 7.71 ± 2.11 mm across five independent infusion runs, outperforming analytical energy-based approaches. To address limited experimental data availability, a conditional generative adversarial network (GAN) is used to adapt synthetic data generated from a digital twin to the experimental domain. When GAN-adapted synthetic data is combined with a limited experimental dataset for training, prediction error is reduced by 46%. These results demonstrate that waveform-based deep learning models, supported by domain-adapted synthetic data, provide a scalable and data-efficient route for accurate resin-flow monitoring in OoA composite manufacturing.
[1A3] Electrical-thermal thresholds and ultrasonic signatures of fault-induced degradation in aerospace carbon fibre-reinforced plastic composites
K M Dikwal¹, V Tunukovic¹, E Mohseni¹, C E Jones¹, K M M Tant², R K W Vithanage¹, C N MacLeod¹, G Munro³, A Douglas³
and S G Pierce¹
¹University of Strathclyde, UK
²James Watt School of Engineering, University of Glasgow, UK
³Boeing Aerospace Innovation Centre (AIC), UK
Electrification in aerospace is advancing primarily in regional, unmanned aerial vehicle (UAV) and electric vertical take-off and landing (eVTOL) platforms to improve efficiency and reduce emissions, while larger aircraft continue to rely on sustainable aviation fuel (SAF), cryogenic fuels and hybrid architectures. This shift involves moving from 28 VDC/115 VAC systems to higher-voltage architectures (±270 VDC to kV-class). A challenge for electrification is weight. While carbon fibre-reinforced polymer (CFRP) structures reduce aircraft mass and could replace metallic casings, their integration introduces new electromechanical failure modes. Uncertainty around mechanical property degradation under electrical conduction remains a critical barrier limiting their use in safety-critical applications. Due to anisotropic conductivity, low-level fault currents can propagate through CFRP, generating localised Joule heating evading conventional protection, and threaten structural integrity. This study evaluates the capability of phased array ultrasonic testing (PAUT) to detect such electrically induced degradation. Quasi-isotropic RTM6 laminates were subjected to controlled currents of up to 7 A, with temperatures tracked relative to glass transition temperature. Robotic PAUT characterised bulk defect signatures, validated through high-resolution micro-computed tomography. The results show high-impedance faults produce localised Joule heating, causing measurable interlaminar shear strength changes before visible damage. A critical electro-thermal delamination threshold is identified and correlated with ultrasonic attenuation and backwall response, demonstrating PAUT’s potential for non-destructive evaluation (NDE) of electrically stressed composites.
[1A4] Prediction of stress distribution around crack tips from ultrasound images using convolutional neural networks
J J An, Q Liu, N O Larrosa, A Velichko and J Zhang
University of Bristol, UK
Structural health monitoring (SHM) is critical for the safety, reliability and long-term performance of engineering structures. This study introduces a machine learning (ML)-driven framework for predicting key fracture parameters, including crack length, stress fields around the crack tip and crack mouth opening displacement (CMOD), directly from ultrasound images. The ultrasonic data were acquired using a permanently installed, low-cost, low-profile ultrasonic array, which provides a much lower signal-to-noise ratio than conventional systems and therefore presents additional challenges for data interpretation. Various types of ML model were built and trained using total focusing method images paired with corresponding stress and strain fields obtained via digital image correlation. This pairing provided explicit crack-related information, enabling predictions of critical fracture metrics. Substantial parameter tuning was undertaken to identify optimal model configurations and explore the relative advantages and trade-offs of various architectures. These comparisons provided valuable insights into how model structure and hyperparameters influence predictive performance. Experimental validation confirmed the reliability of this framework. By integrating low-cost ultrasonic imaging and ML, this work underscores the potential of leveraging data-driven techniques to transform SHM practices. The findings pave the way for scalable, cost-effective and computationally efficient solutions to monitor structural integrity in real time, offering significant benefits for infrastructure management and maintenance.
Keywords: ultrasonic arrays, digital image correlation, crack length, defect characterisation, structural health monitoring.
[1C1] Using system performance verification to reinforce safety, quality and reliability in NDT operations
P Hillman and J Poirier
Eddyfi Technologies, Canada
In non-destructive testing, protecting the public and the profession depends not only on compliance with standards and certification requirements but also on data integrity captured in the field. Inspection teams are increasingly expected to maintain productivity in demanding environments while preserving confidence in instrument performance and data quality.
This paper presents a system performance verification (SPV) approach that provides structured, repeatable and documented checks of key instrument functions without requiring return to an original equipment manufacturer (OEM) service centre. Using a traceable verification device and guided software workflow, the method runs an automated sequence confirming whether the system is operating within expected performance limits.
The paper will describe how SPV is performed, what aspects of instrument behaviour are evaluated and how the results increase confidence in equipment readiness. It also clarifies the distinction between verification and formal accredited calibration, emphasising that SPV complements rather than replaces required certification processes.
The paper further examines how SPV supports stronger quality practices through increased equipment checks, improved traceability and better escalation decisions when instrument behaviour appears abnormal. Ethical and human factors related to equipment uncertainty in high-pressure inspection environments are also considered.
[1C2] Subsea DRT through concrete coating: validation and calibration study
J McNab and J Moore
Oceaneering International Services Ltd, UK
Inspection of non-piggable concrete-coated subsea pipelines presents significant challenges for non-intrusive inspection (NII) in comparison to the more common internal line inspection (ILI) methods. Installing pig launchers and receivers is often not possible, whereby a lot of old pipelines have bends <5D where a pig would normally need a 5D bend to pass.
Current NII methods include marinised pulsed eddy current (PEC) and more lately acoustic resonance technology (ART), but have limitations with respect to flaw sizing due to the heterogeneous nature of concrete weight coating and the presence of embedded reinforcement, in addition to a usually large probe ‘lift-off’ due to the concrete thickness. More traditional inspection methods often require coating removal, production interruption or invasive intervention.
This paper presents the validation and calibration of a digital radiography technology (DRT) methodology developed for the inspection of a 24" subsea gas export pipeline operated by a major oil and gas provider.
An onshore validation programme was conducted using a representative test spool incorporating artificial corrosion and crack-like defects. Results demonstrated reliable detection of corrosion-type wall loss through 57 mm concrete coating, successful identification of crack-like features and the establishment of a repeatable calibration relationship between radiographic greyscale response and remaining wall thickness.
The work provides a technical foundation for offshore deployment of subsea digital radiography in concrete-coated pipeline applications.
[1C3] High-temperature DC conductivity methods for non-standard samples
R Howles
National Physical Laboratory (NPL), UK
Direct current (DC) conductivity is a critical material property in the aerospace, automotive and energy sectors, informing design choices for components from engine parts to structural steels. The drive to electrification demands DC conductivity data representative of operational conditions, including elevated temperatures. Four-terminal techniques for measuring DC conductivity are well established and involve simple assemblies that can be adapted for high-temperature measurements. In response to a lack of industrial traceability in DC resistivity for aerospace, the National Physical Laboratory (NPL) built a traceability chain to establish robust estimated uncertainties and optimise the methods, leading to United Kingdom Accreditation Service (UKAS) accreditation in 2024. Since then, growing industrial needs have meant the broadening of standard techniques to accommodate higher temperatures, non-standard sample geometries and novel materials. Two recent case studies underlining these developments will be presented. The first is an adaptation of the knife edges set-up for small permanent magnet samples measured up to 200°C in varied states of magnetisation. The second is an adaptation of the van der Pauw method for inhomogeneous thin-film composite samples. Together, these developments demonstrate practical routes for extending traceable, non-destructive DC conductivity measurements to industrially relevant materials and geometries that are not well served by standard methods.
[2A4] High-frequency electromagnetic tomography (EMT) sensor for the monitoring of saline diffusion in porous media
X Zou, Y Li, C Jiang, S She, X Pang, A Peyton and W Yin
University of Manchester, UK
Porous materials have received increasing attention due to their widespread involvement in environmental and engineering applications. Salinisation phenomena in such materials are commonly encountered in soil science and wood processing, highlighting the need for effective inspection of saline diffusion processes.
Electromagnetic tomography (EMT) is an emerging imaging modality that maps the cross-sectional electromagnetic distributions. However, most EMT research up to date focuses on high-conductivity metals using low-frequency fields, which are less effective for monitoring low-conductivity materials such as saline, and totally ineffective for many non-conductive porous materials.
To address this limitation, this study develops a high-frequency EMT sensing system that achieves an SNR of 76.74 dB, operating at frequencies up to 5 MHz. The system performance is initially validated through imaging bulk liquid samples, including salines from 0.25 % to 20 %. The study then extends to monitoring porous media where packed structures formed by 5 mm and 8 mm glass beads immersed in 4 % saline are imaged. Additionally, dynamic diffusions of 4 % saline within sponge, soil and wood slice are captured for the first time based on this technology. Results demonstrate that the high-frequency EMT sensing system provides a novel and effective tool for the monitoring of dynamic saline diffusion in porous materials.
[2A5] Automated inspection and reporting of industrial weld coupons
A Gilmour¹,², W Jackson¹, G Dobie¹ and C MacLeod¹
¹University of Strathclyde, UK
²BAE Systems, UK
The assessment of welder competency relies on the manual inspection of welded test coupons that are produced to demonstrate a welder’s ability to create sound welded joints. Ultrasonic non-destructive testing techniques, including phased array ultrasonic testing and time-of-flight diffraction, provide a reliable means of detecting and characterising internal weld defects while preserving the integrity of the test-piece. However, such tasks across a wider manufacturing environment can be deemed lower priority to inspectors as they take individuals away from more critical jobs. Current inspection procedures typically rely on manual measurements and calibration, inducing longer set-up times.
This work investigates the automated inspection of weld coupons to remove the burden of set-up and collection from the inspector. Effective automated inspection requires accurate localisation of the weld to ensure the ultrasonic probes are positioned correctly relative to the joint. Although currently achieved through measurements of external references, the research also aims to explore the feasibility of using the ultrasonic data to identify the weld position and guide probe placement during scanning. This approach aims to reduce reliance on manual measurements, improve inspection repeatability and support the development of more automated inspection workflows.
[2A6] Automated robotic magnetic Barkhausen noise measurement via sensor-driven pose and contact control
X Pang, S She and W Yin
University of Manchester, UK
Magnetic Barkhausen noise (MBN) is sensitive to stress and microstructure, but its use in robotic inspection is limited by probe orientation, lift-off and contact variability. A hierarchical sensor-driven method is presented for automatic MBN measurement on planar ferromagnetic surfaces where inclination is not known in advance. A triangular three-coil eddy current head first provides non-contact estimates of two angular errors and the surface-normal offset. After robot alignment, a calibrated tool transformation places the MBN probe over the inspection region. Contact is then established by a coarse-to-fine approach in which an MBN root-mean-square (RMS) threshold triggers fine motion and consecutive RMS differences define convergence. The combined procedure was evaluated from 0° to 60° in 10° increments, with ten 2 s acquisitions at each inclination. Signals were sampled at 100 kS/s under 25 Hz magnetisation. For each acquisition, the Hilbert envelope of the hardware-filtered MBN voltage was paired sample by sample with the synchronously recorded H signal to construct the H-MBN feature curve. Eight parameters describing curve amplitude, peak position and width were used to assess stability. Across 70 measurements, their pooled within-angle coefficients of variation were 0.245-2.957%, while the relative ranges of the seven angle means were 0.405-2.005%. The results demonstrate repeatable end-to-end measurement over a wide inclination range without a prior surface model or force/torque feedback.
[2C1] DHAMES: smart magnetic NDT system. A new concept and instrument for NDE of quality of heat-treated ferromagnetic steel components
M Vaidhianathasamy
SAVI3M Tech UK Ltd, UK
The DHAMES system is a new and first-of-its-kind instrument for non-destructive evaluation of material quality of heat-treated ferromagnetic steel components such as shafts, pistons, gears and bearings, etc. The DHAMES system uses the unique concept of measurement and analysis of distortion and harmonic levels of the magnetic excitation voltage signal across the coil around the electromagnetic yoke that is driving the cyclic magnetisation of ferromagnetic steel components. Another unique aspect of the DHAMES system is that it just uses a single bespoke electromagnetic yoke and there are no other pick-up coils or sensors used. The DHAMES measurements involve electromagnetic excitation of the ferromagnetic steel component using a bespoke electromagnetic yoke designed specific to the geometry of the components such as a gear flank and shaft surface curvature. The analysis of the distortion to the excitation voltage signal (Ve) is carried out by the calculation of time derivative of the excitation voltage profile (dVe/dt) and the changes in the levels of frequency harmonics in response to the magnetisation of the ferromagnetic material under inspection. The DHAMES measurements can be used to identify variations in the microstructure, hardness and case depth, etc, of heat-treated ferromagnetic steel components by measuring directly on the surface of the components. This paper presents the applications of the DHAMES system for the evaluation of hardening treatment, for the determination of case depths in induction-hardened ball shaft pins and carburised gear flanks and for the evaluation of compound layer thickness in nitrocarburised pistons.
[2C2] Passive muon imaging as a dual-purpose inspection and monitoring tool for industrial assets
N Zabari, K Katrankova, M Dobrowols and R Bożek
Muotech, Poland
Cosmic-ray muon imaging is emerging as a powerful non-destructive technique for inspecting structures that are inaccessible to conventional NDT methods. However, most industrial applications reported to date have focused on static imaging. This work demonstrates that muography can provide not only structural information but also direct insight into the operational state of large industrial assets during normal operation.
A measurement campaign was conducted using a portable muon tracking system deployed at three locations around a large operating industrial vessel. More than 30 million muon trajectories were recorded without interrupting operations or requiring access to the asset interior. Independent measurements from multiple viewing angles consistently resolved previously inaccessible internal structural features, demonstrating the robustness of the technique under field conditions.
Beyond the primary imaging objective, the measurements revealed a strong and repeatable time-dependent modulation of the muon signal caused by changes in the internal mass distribution of the asset during routine operation. This observation demonstrates that passive muon measurements can track dynamic density changes in real time while simultaneously providing structural information.
The results suggest a significant expansion of the role of industrial muon imaging: from a niche imaging technique to a practical tool capable of combining non-intrusive inspection, structural verification and continuous operational monitoring within a single measurement system. The paper will present the deployment strategy, the reconstruction approach, the multi-view imaging results and the implications of dynamic muon monitoring for future industrial NDT applications.
[2C4] Phased array ultrasonic testing of socket welds
R Jenkins
Uniper Technologies Ltd, UK
A significant proportion of steam leaks in gas power generation systems occur due to socket weld failures. Volumetric non-destructive testing (NDT) is challenging due to the limited access, challenging geometry and the small dimensions involved. Advanced phased array ultrasonic testing techniques were trialled on socket weld samples with manufactured flaws. All crack-type flaws were detected. In total, 32 of 34 flaws were detected. Mean height sizing was within ±1 mm. A validated technique was developed and deployed on site. Further validation work was planned to increase the dataset.
Plenary Paper: The evolving landscape of NDT in the UK
T Barden
Rolls-Royce plc, UK
There are many influences creating change in the non-destructive testing (NDT) industry, including emerging technologies, automation, new materials, novel applications and workforce dynamics. These developments present risks as well as opportunities to deliver greater value for the UK economy. Over the last year, the British Institute of Non-Destructive Testing (BINDT), Innovate UK Business Connect and the UK Research Centre in NDE (RCNDE) have run workshops and held a survey to gain perspectives from industry, researchers and stakeholders on how NDT may develop in the next five to ten years.
Building on the 2014 report ‘A landscape for the future of NDT in the UK economy’, this study aims to understand how the UK NDT industry should evolve over the coming decade and what capabilities and actions are required to enable the transition.
[3A1] CAD-free, fixtureless automated immersion inspection through ultrasound-driven surface adaptation with a linear phased array
R Moos
University of Strathclyde, UK
The precise nature of non-destructive testing (NDT) combined with the variable nature of the parts that need to be inspected has limited the adoption of NDT automation. Often, the meticulous fixturing, need for a computer-aided design (CAD) model or time-consuming point-to-point programming render the process for accurate automated inspection economically untenable. Sensor-enabled automation using devices such as cameras or force detectors has been proposed to close this economic gap. However, these sensors have limited capability in immersion and they ignore an obvious solution: the NDT data already acquired. This paper explores the adoption of a 1D linear phased array probe as the sensor to maintain a surface-normal orientation and fixed stand-off distance relative to the component. Live ultrasound data was combined with known robotic positional data to enable real-time surface adaptation, which simultaneously facilitated the production of an accurate volumetric reconstruction of the entire component, including internal defects. In this study, a multi-angle stepped testing block and a representative spoiler section (both aluminium) were utilised to evaluate the effectiveness of the adaptive compensation against reference reflectors. This research showed that optimal speed and inspection accuracy were achieved using ultrasound driven path planning. Furthermore, this integration of NDT and robotic programming extends naturally towards applications including automated repair and position compensation for complex surface geometries.
[3A2] Robo-SRAS: in-situ material characterisation of arbitrary geometries using laser ultrasonics
M Todd, R Patel, W Li, R J Smith and M Clark
University of Nottingham, UK
Spatially resolved acoustic spectroscopy (SRAS) is a laser-ultrasonic technique capable of imaging microstructure through measurements of surface acoustic wave velocity. To date, SRAS has primarily been confined to laboratory-based systems operating on planar specimens using precision translation stages. This paper presents RoboSRAS, a compact robotically deployed SRAS system designed to extend microstructural inspection to geometrically complex surfaces. The system combines fibre-delivered ultrasonic generation and a commercially available rough-surface laser-ultrasonic detector within a lightweight measurement head mounted on a six-axis robotic arm. Characterisation on an acoustically homogeneous reference specimen showed that waveform averaging reduced the median absolute deviation in measured surface acoustic wave velocity from 75.7 m/s to 29.8 m/s, while maintaining agreement with measurements obtained using a conventional laboratory SRAS system. Measurements on a machined coarse-grained titanium specimen successfully resolved microstructural contrast corresponding to the underlying grain structure, demonstrating operation on industrially relevant surface conditions. The work demonstrates the feasibility of robotically deployed microstructural imaging using laser ultrasonics and provides a path towards rapid, non-contact and non-destructive material state characterisation of complex engineering components without the use of couplants or destructive sample preparation.
[3B1] Digital twins for industrial X-ray and CT inspection systems: an asset administration shell approach
D Beck and F Herold
VisiConsult X-ray & Solutions GmbH, Germany
As machines, including X-ray and computed tomography (CT) systems, get more and more connected in the context of Industry 4.0, it is increasingly important to have a digital representation of physical assets, a so-called digital twin. To enable the usage and exchange of digital twins even across organisational borders, a unified standard is required. For this reason, the Industrial Digital Twin Association (IDTA) has developed the Asset Administration Shell (AAS), providing a standardised way to structure digital twins. VisiConsult is developing a digital machine model based on the AAS that aims to provide a unified representation for any industrial X-ray and CT inspection machine, covering aspects such as detector and source specifications, axis configurations, supported trajectories and more. This general representation allows possible applications to be evaluated ahead of time and will simplify the creation of new software tools that require knowledge of the machine specification. This talk will present the AAS model structure we arrived at and implementation challenges we faced on our way (and are still facing). Finally, we will discuss how we can in future create value for customers adopting the AAS for inspection machines or inspected products.
[3B2] Inspection 4.0: unlocking AM design freedom
T Maw, W Vesga, H Leach, J Chamberlin and H Greenhalgh
The Manufacturing Technology Centre (MTC) Ltd, UK
Additive manufacturing (AM) enables complex geometries, lattice structures and integrated thermal management features that are difficult or impractical to manufacture using conventional processes. However, the ability to exploit this design freedom remains constrained by verification capability. In many AM workflows, inspection remains dependent on post-build methods that are time-consuming, manually interpreted and weakly connected to the original design intent. This creates a practical gap between what can be built and what can be confidently accepted.
This paper describes a model-based enterprise (MBE) inspection workflow developed to connect design intent, in-process eddy current (EC) inspection and post-build X-ray computed tomography (XCT) reference measurement for additively manufactured artefacts. The work focuses on evaluation of seeded cylindrical defects because they provide a controlled basis for demonstrating the digital thread through model-based definition (MBD), automated dimensional XCT evaluation and manual in-process EC analysis. The central purpose is to present how in-process EC indications and XCT reference measurements can be connected to common feature identities and coordinate frames, which enables comparison and future in-process inspection validation efforts.
Test artefacts containing 51 intentionally seeded cylindrical defects per test-piece were designed with full MBD as machine-readable authority product specification data. The MBD was used throughout manufacturing and inspection processes, connecting the design specification to the physical build, in-process EC inspection and post-process XCT inspection. This approach is shown to have significant advantages over current state practices in the AM process chain, particularly in the measurement and inspection domains, where it enables automatic evaluation of geometric dimensioning and tolerancing (GD&T) and provides the required coordinate reference frames and data traceability to compare in-process and post-process inspection results for key defect characteristics: size, location, form and orientation.
The results show clear EC indications for the analysed seeded defects and demonstrate a practical route for comparing EC responses with XCT-derived reference measurements. The illustrative EC/XCT comparison showed closer agreement in vertical sizes than in horizontal sizes or positions. The known reasons for this include EC probe diameter, probe-response broadening, point-spread behaviour and coordinate or scaling calibration challenges. The work therefore demonstrates a pathway towards in-process verification rather than a fully validated replacement for post-build inspection. The central conclusion is that the challenge extends beyond sensing capability alone: traceability, coordinate management, characteristic naming and automation are required to turn in-process signals into verification evidence.
[3B3] Beyond the core: advanced ECA inspection solutions in nuclear power plants
C Tremblay, S Loffredo, M Bouchard and L Hallal
Eddyfi Technologies, Canada
The revival of nuclear energy is a key part of the global energy transition aimed at reducing carbon emissions and addressing climate challenges. However, sustainable nuclear development depends on maintaining the highest safety standards, which remain the foundation of the industry’s legitimacy. Beyond technical compliance, safety requires a strong organisational culture, effective prevention, incident management systems and transparent communication to maintain public trust.
Non-destructive testing (NDT) plays a critical role in supporting nuclear safety by enabling early defect detection, integrity assessment and failure prevention throughout manufacturing, construction, operation and maintenance phases. Among NDT methods, eddy current testing, particularly eddy current array (ECA) technology, is increasingly recognised as a reliable alternative to conventional surface inspection techniques, which typically have a limited area of coverage. The use of ECA improves inspection quality, repeatability, traceability and probability of detection while reducing operator dependency. This technique is also well suited for remote deployment by robotics.
To maximise effectiveness, these technologies require qualification, standardisation and regulatory recognition. Eddyfi Technologies actively collaborates with inspection companies, nuclear operators and institutions to deliver innovative eddy current solutions for applications such as fuel rods, steam generator tubes, control rod drive mechanism (CRDM) J-welds and nuclear waste container inspection, supporting the industry’s highest safety standards.
[3B4] A review of surface eddy current array use in industry
J M Watson
TWI Ltd, UK
Eddy current array (ECA) technologies are frequently promoted as a replacement for conventional surface inspection methods, including magnetic particle and dye penetrant. In recent years, interest in their industrial adoption and commercial availability has increased significantly.
This paper reviews the application of ECA technologies across a range of industrial case studies, examining both successful implementations and instances in which technical limitations restricted their effectiveness. The underlying technical justifications for each application are discussed, highlighting the factors that influenced inspection performance or the lessons learned.
This paper also provides a critical assessment of ECA implementation in industry, identifying principal barriers to wider acceptance and general use.
[3C1] Damage-index-guided time-frequency 2D-CNN learning for internal defect detection in titanium alloy using ultrasonic A-scan signals
Z Qin¹, H Liu¹, P Li¹, B Wang¹, D Liu² and Z Qian¹
¹Nanjing University of Aeronautics and Astronautics, China
²University of East Anglia, UK
Ultrasonic non-destructive testing for titanium alloy components faces difficulties due to severe material grain noises in A-scan signals, making it difficult to detect tiny internal defects, which greatly limits the stability of detection. A damage-index-guided time-frequency learning procedure is proposed for detecting internal defects in titanium alloy members using time-frequency components of ultrasonic A-scan signals. The pulse duration is estimated using threshold-, envelope- and frequency-based criteria after temporal alignment of the healthy reference and inspected signals. A local damage index is calculated via a moving window strategy to identify the signal region most sensitive to defect-induced variations. The selected segment is transformed into a time-frequency representation by the continuous wavelet transform (CWT) and the resulting spectrogram is classified by a 2D convolutional neural network (2D-CNN). By restricting the time-frequency analysis to defect-relevant regions, the proposed approach suppresses interference from irrelevant signal components while preserving transient and spectral characteristics associated with shallow defects. This framework effectively integrates interpretable damage-index analysis with deep feature learning, thereby enhancing the reliability and robustness of ultrasonic inspection for titanium alloy components.
[3C2] Characterisation of polycrystalline materials through ultrasonic imaging
I Aizpurua-Maestre¹, J L Lanzagorta¹, U Arregi¹, U Iparragirre¹, I Sanchez¹, J Mendikute¹ and L Galdos²
¹IDEKO – Basque Research and Technology Alliance, Spain
²Faculty of Engineering of Mondragon Unibertsitatea, Spain
The ultrasonic response of polycrystalline materials is strongly influenced by their microstructural characteristics, particularly in the presence of large and anisotropic grains. Under such conditions, ultrasonic wave propagation is affected by phenomena such as scattering, beam distortion and attenuation, which complicate the interpretation of inspection results. Although advanced ultrasonic imaging techniques have been proposed to address these challenges, their interaction with complex microstructures and their underlying physical behaviour are not yet fully understood.
In this work, post-processing strategies based on ultrasonic imaging concepts are developed to analyse and characterise the influence of microstructural heterogeneity on the ultrasonic response of polycrystalline materials. An austenitic alloy with a heterogeneous grain structure is investigated through a combination of numerical simulations and experimental ultrasonic inspections. Grain size variability is independently assessed by microscopy analyses, enabling a direct correlation between microstructural features and observed ultrasonic responses. The study provides insight into the relationship between microstructure and ultrasonic signal behaviour, contributing to a better understanding of inspection limitations in coarse-grained materials.
[3C4] Radio-transparent turntables in digital radiography: optimising access for complex aerospace components
J Deacon
GKN Aerospace Filton, UK
Real-time radiographic inspection of complex aerospace components is frequently compromised by geometric constraints, where traditional turntables and tooling obstruct critical regions of interest. This paper presents the implementation of a novel digital radiography configuration utilising a large-scale radio-transparent turntable. By eliminating the radio-opaque fixtures that are standard in conventional systems, this set-up enables near 360° manipulation without bespoke tooling, significantly reducing the number of required axes. Validation using standard wire-type image quality indicators (IQIs) and duplex gauges demonstrated full compliance with ISO 17636-2 Class B (Compensation Principle II) and Nadcap requirements. The results show that this approach not only resolves geometric access issues but also optimises the inspection cycle time while maintaining high-sensitivity defect detection.
[4A1] Feature segmentation for ultrasonic non-destructive testing of carbon fibre-reinforced polymers
M Kwasnik¹, V Tunukovic¹, E Mohseni¹, S G Pierce¹, A Hamilton², R K W Vithanage¹, C N MacLeod¹, G Munro³ and C Smith³
¹University of Strathclyde, UK
²Digital Process Manufacturing Centre (DPMC), UK
³Aerospace Innovation Centre (AIC), UK
Carbon fibre-reinforced polymers (CFRPs) used in the aerospace sector require thorough non-destructive testing (NDT) to ensure structural integrity. Phased array ultrasonic testing (PAUT) is highly effective in reducing inspection time; however, it creates a bottleneck in the workflow by generating high volumes of data. Additionally, variability in CFRP component geometries, manufacturing processes and inspection conditions complicates automation of data interpretation.
This work investigates vision foundation models (VFMs), which leverage extensive pre-training, to provide improved generalisation across variable data. VFMs are fine-tuned in a supervised training approach on a diverse corpus of PAUT B-scan and D-scan data to perform pixel-wise segmentation of features such as defects and structural echoes (front/back walls). Structural features enable estimation of component geometry, supporting automated defect characterisation.
Data in this work was acquired with a KUKA KR90 manipulator delivering PAUT inspection of industrial-grade CFRP components. The initial results on an experimental sample containing 25 flat-bottomed holes show all defects successfully detected and with a mean Dice score of 80%, measuring overlap between true and predicted defect regions. Ongoing development introduces segmentation of structural features to improve feature discrimination, enhance defect characterisation and demonstrate the broader potential of this approach for automated CFRP inspection.
[4A2] Extraction of metallic material microstructure grain size from total focusing ultrasonic images
J Zhang
University of Bristol, UK
Ultrasonic backscattering signals from material microstructures offer a way to assess grain size within the material. Typically, this process uses a focused ultrasonic transducer to perform pulse-echo immersion measurements at various spots, aiming to calculate the root-mean-square amplitude of the backscattered signal at a chosen focal point for a reliable grain size estimate. However, this approach limits its use in practical scenarios, such as real-time monitoring in high-value manufacturing or in-service evaluations, where repeated immersion measurements are not practical. The key advantage of ultrasonic phased arrays lies in their ability to focus ultrasound beams at multiple locations from a single probe position, using different focal laws applied either directly or during data post-processing. This capability suggests that precise grain size measurements could be achieved with just one array measurement. In this study, the traditional backscattering technique, originally developed for standard transducers, is modified to work with full matrix capture (FMC) data from an ultrasonic array to provide metallic material microstructure measurement of grain size.
This paper develops an inverse model to determine material grain size using a one-dimensional (1D) array in a direct contact set-up, adapting the traditional backscattering model for array-based inspections. The model calculates wave pressure using Huygens’ principle. The application of this model to experimentally collected FMC array datasets and total focusing ultrasonic images is demonstrated. In this approach, grain size is assessed by simultaneously leveraging backscattering amplitudes and material attenuation. The model assumes uniform grain size statistics and consistent material attenuation across the entire microstructure.
Tests were conducted on steel specimens to validate the approach. The grain sizes derived from array data closely match those obtained through metallography, demonstrating strong consistency. Unlike traditional pulse-echo immersion backscattering techniques, this method accurately measures grain size using a direct contact set-up with fewer measurement points, offering the potential for real-time grain size monitoring.
[4A3] Using dual linear array transmission-receive longitudinal (DLA TRL) to inspect thin-walled stainless steel nuclear assets
S Serjeant, C MacLeod and E Mohseni
University of Strathclyde, UK
Sellafield has been the cornerstone of the UK’s nuclear power generation due to the site’s fuel reprocessing facilities. In recent times, economic and political shifts have resulted in the closure of the UK’s only commercial-scale reprocessing facility. This shift has created the unique problem of repackaging the previous generation of short-term canisters into a package more suited for long-term storage. Utilising autogenous gas tungsten arc welding (AGTAW), the new generation of canisters must meet strict control measures. Current inspection methods require acceptance sampling, whereby an individual canister from a batch is sent off for destructive testing and inspection, determining the fate of the batch. The key defining factor of a defective canister is failing to meet the tight thickness requirements between the surface and the canister’s unique lack of fusion features present. Canisters must meet a minimum thickness of 2.1 mm. To assess that each canister meets the requirements, dual linear array transmission-receive longitudinal (DLA TRL) presents the solution to providing a truly non-destructive evaluation of every canister before deployment. Typical phased array ultrasonic testing (PAUT) is sufficient when considering thick geometry; with a 3 mm nominal thickness across the canister body, the potential for the front wall to obscure the reflection from a near-surface reflector is highly probable. To counter this, DLA TRL presents a promising solution. By utilising CIVA beam computations in conjunction with immersion DLA TRL, early-stage prototyping of a tailored TRL wedge can be achieved. This initial phase of research presents a small selection of probable designs, with a final design being narrowed down. The final solution excels where traditional PAUT does not, by removing front wall reflections and mechanically focusing energy into the required region, improving defect visibility and providing a sufficient baseline for future endeavours.
[4B1] Diamond quantum sensors for non-destructive testing: application to additive manufacturing
B Vindolet, G Bourcin, H N Nguyen and T Hingant
Kwan-tek, France
A novel technique is presented for non-destructive evaluation (NDE) based on the nitrogen-vacancy (NV) centre in diamond. NV centres enable the detection of defects in magnetic and conductive materials at or near the surface, by simultaneously measuring the magnetic leakage field and the eddy current signal. This diamond quantum sensor combines high sensitivity with high resolution in a small and ready-to-use endoscopic probe. NV-based NDE could offer a way to improve existing electromagnetic NDE by detecting micrometre-scale defects currently difficult to observe, by increasing the measurement depth of eddy current devices and by deploying NDE in harsh environments and constraint spaces.
This presentation will focus on new experimental results obtained in industrial use cases of additive manufacturing. It is indeed useful for these applications to detect micrometre-scale porosity defects across multiple deposited layers, ie over a depth of a few millimetres. Ultimately, we plan to integrate NV sensors into the additive manufacturing device to enable real-time defect monitoring. We will take this example to illustrate the potential interest of NV-based NDE and to present our prototype, integrating all our R&D measurement set-up into a compact device enabling our partners to use our NV technology in their own environment.
[4B2] Eddy current testing for non-destructive inspection of carbon fibre-reinforced polymer multi-strand cables
F Bolinhas¹, T G Santos¹, M A Machado¹, J de la Cuesta² and S Cooley²
¹Universidade NOVA de Lisboa, Portugal
²Future Fibres Rigging Systems SL, Spain
Multi-strand cables built from pultruded carbon fibre-reinforced polymer (CFRP) rods and protected by a braided textile jacket are progressively replacing steel in weight-critical structural applications such as yacht rigging, bridge stays, tidal turbine tethers and large scientific instruments. Assessing the condition of these components in service remains an open problem: ultrasonic testing, radiography and off-the-shelf eddy current equipment all perform poorly when confronted with the strong electrical anisotropy of the composite, the curved multi-strand geometry and the presence of a textile cover that conceals strand level damage. In this work, a purpose-built eddy current inspection solution is presented. Finite element modelling was employed to compare candidate coil arrangements and to select the geometry offering the greatest sensitivity to rod discontinuities, after which the chosen design was manufactured on a printed circuit board substrate. The resulting probe combines a single excitation coil with a pair of reading coils wired in a differential arrangement; the reading coils are oriented tangentially to the cable surface, with the excitation coil placed perpendicular to them. Complete coverage of the cable circumference was obtained by combining longitudinal sweeps, executed by a collaborative robot, with successive angular increments provided by a rotational fixture, and a hand-held variant of the equipment was additionally produced for field use. Deliberately introduced damage of several morphologies, comprising crushed, severed and separated rods in the two outermost layers, was identified through the sheath without removing it. Purpose-written software automates the acquisition and converts the measurements into two-dimensional C-scan maps and a three-dimensional rendering of the results over the cylindrical geometry. The indications obtained matched the position and extent of the physical damage, confirming that tailored eddy current testing constitutes a viable route for the in-service evaluation of multi-strand CFRP cables.
[4B3] Development of robotic eddy current array inspections for complex geometries
A D’Amours, S Loffredo, S Grond and M Bouchard
Eddyfi Technologies, Canada
Advances in eddy current array (ECA) technology have made it possible to expand its applicability for increasingly complex geometries, such as dovetail assemblies in aerospace turbines as well as turbine blades. ECA is commonly adopted as a replacement for manual techniques in the detection of surface defects, as it reduces operator dependency whilst enabling permanent storage of inspection data. Additionally, inspection speed is significantly higher, optimising the deployment of inspectors and analysts.
Integration of ECA inspection into robotic systems for production of complex, high-value components reduces limitations inherent to manual inspection, including variations in pressure applied by the operator and trajectory control. To facilitate deployment of a robotic system, a collaborative robot (cobot) can initially be used, as this offers several advantages. Cobots have safety features that enable continuous human/machine collaboration and programming a cobot’s trajectory requires no coding, allowing continuous improvement; once optimised, it becomes possible to deploy a fully automated system using an industrial robot.
This paper demonstrates how ECA technology can be integrated into a robotic system, from the selection of eddy current sensors to iterative trajectory programming strategy. This solution represents a significant gain in terms of inspection reliability compared with conventional techniques or manual ECA.
[4C1] A revolutionary ultrasonic thickness gauge: PM10 from Yushi Instruments, China
A Liu¹, M Soare² and M Valentin Predoi³
¹Yushi Instruments, China
²Sunrise Materia Research & Innovation (Suma Rei), Romania
³Politehnica University of Bucharest, Romania
Suma Rei specialises in market studies of high-performance non-destructive testing (NDT) instrumentation and accessories. The company performed extended marketing studies on the high accuracy of ultrasonic testing (UT) thickness instruments and found an exceptional ultrasonic thickness gauge and specific accessories produced by Yushi Instruments, a company based in China. Yushi Instruments is a powerful, high-tech enterprise focused on the research and development (R&D) and manufacturing of NDT equipment and the company is committed to independent innovation in its field of activity. This collaboration offers customers a high-resolution and affordable C-scan machine coupled with the Yushi thickness measuring device, resulting in an unbeatable performance system.
[4C3] Virtual-reality-based programming-by-demonstration system for automated robotic ultrasonic inspection
S Ramezani Talabari¹, R K W Vithanage¹, S G Pierce¹, E Mohseni¹, K-H Lam², G Munro³ and T O’Hare⁴
¹University of Strathclyde, UK
²University of Glasgow, UK
³Boeing Aerospace Innovation Centre, Glasgow Prestwick Airport, UK
⁴Short Brothers, a Boeing Company, UK
Robotic ultrasonic testing promises consistent, repeatable inspection of complex components, yet conventional online and offline programming methods are generally expensive, time-consuming and demand dual expertise in robotics and non-destructive testing, creating a barrier to adoption. This study presents a programming-by-demonstration method that addresses these limitations using a consumer-grade virtual reality tracking system, with which an operator demonstrates the inspection path directly on the workpiece using a hand-held teaching device. Demonstrated poses are resolved into the robot base frame through a continuously updated reference-tracker registration and then refined by a multi-stage post-processing pipeline that enforces probe normality to the surface and roller-direction alignment, rendering the demonstrated motion executable by the robot. The corrected trajectories are executed under hybrid position-force control with synchronised phased array data acquisition. Experimental validation on a curved aluminium specimen with flat-bottomed hole reflectors across five workspace placements yielded a mean programming time of 15 s, compared with 21 min 28 s for teach pendant programming, a reduction of 98.8%, whilst detecting reference reflectors with a signal-to-noise ratio above 17.2 dB, a mean lateral positional error of 3.02 mm and a mean depth error of 0.59 mm. The dynamic registration was validated by relocating the robot within the workspace and by repositioning the tracking base stations: in both cases, the system executed the inspection without recalibration, confirming its suitability.
[5B1] Advances in X-ray CT of large and dense AM components
N Brierley, K Mrzljak, M von Schmid, M Weiß, B Zengerling and O Günnewig
Diondo GmbH, Germany
The rise of large-format laser powder bed fusion additive manufacturing (LPBF-AM) for applications such as space flight, often using dense materials such as copper or nickel alloys, presents a significant inspection challenge. Given the geometric complexity of samples that are typically manufactured by this route, and range of potential quality issues to be detected, X-ray computed tomography (CT) has established itself as the default post-build inspection method. However, the scale of the samples now becoming manufacturable, in combination with the density, and hence X-ray absorption, of materials being used, requires the bounds of CT to be pushed. For example, even a 450 kVp source may not provide the necessary penetration and so MeV imaging, using a linear accelerator (linac), becomes necessary. A selection of advanced techniques are presented that Diondo is exploring to improve the imaging capabilities of CT technology in this context, including algorithmic means of increasing detector resolution and optimised use of linac pulses, as well as the use of next-generation detectors.
[5B2] Evaluation of flexible digital detector arrays against the ISO and ASME weld radiography standards
S Burch and L Zhao
ESR Technology Ltd, UK
Digital radiography using rigid flat-panel digital detector arrays (DDAs) and flexible imaging plates (IPs) has been in widespread use for many years. However, flexible digital X-ray detectors have become commercially available only recently. For filmless radiography needing curved detectors, for example in the double-wall single image technique, these flexible arrays provide an alternative to IPs and potentially provide significant advantages in terms of image quality and sensitivity.
However, the characteristics of these new detectors have not been independently assessed to establish how effective they are compared with IPs and whether they are compliant with the requirements of international standards such as ISO 17636-2 and ASME V Article 2 for manufacturing weld radiography.
ESR organised a HOIS trial in 2025 to evaluate flexible digital detectors using three different types of radiation source: X-ray, Se-75 and Ir-192. Three suppliers of flexible digital detector arrays participated in the trial. This paper describes how the trial was conducted and how the results were analysed against the requirements of ISO 17636-2 and ASME V Article 2.
[5B3] A comparison study of linear stages and robotic arms for digital tomosynthesis
L Eley
University of Liverpool, UK
Computed laminography (CL) is an X-ray imaging modality that uses several projections of an X-ray source through an object to create a 3D reconstruction. Previous work has shown the benefit of this approach compared to competing modalities and discussed how to optimise the arrangement of biaxial X-ray projections to maximise the quality of the resultant 3D reconstruction. In this study, two CL imaging systems are compared. The image quality obtained from a prototype scanner consisting of a lightweight X-ray source mounted on a UR3e robotic arm was benchmarked against existing biaxial systems that use linear translation stages. Standard image quality metrics are used to quantify this, with the conclusions drawn supported by high-quality images of some typical use cases relevant to non-destructive testing and evaluation (NDT&E) assessments. From this combination of work, no quantifiable advantage is found to using a three-axis ‘triaxial’ approach enabled by the cobot over the current biaxial product solution. The importance of scan angle in improving reconstructed image quality is emphasised and compared to other less impactful factors, including primarily the source emission position geometries and source-to-image distances.
For further information contact: Events and Awards Department, The British Institute of Non-Destructive Testing,
Midsummer House, Riverside Way, Bedford Road, Northampton NN1 5NX, UK. Tel: +44 (0)1604 438300; Email: conf@bindt.org
