This project is funded by the International Science Partnerships Fund (ISPF) under the International Collaboration Awards 2024 – Brazil and South Africa scheme.

Northumbria University (UK) coordinates the overall technical delivery and integration, leading the development of thermography-based inspection capability for curved composite structures and the associated reconstruction and interpretation workflow. Northumbria also drives training and knowledge exchange for early-career researchers across the programme.

Federal University of Rio de Janeiro (Brazil) leads Brazil-side experimental development and application-oriented validation, including heat-transfer modelling support and practical inspection trials on relevant blade components and curved industrial structures. UFRJ also enables local engagement and pathways for adoption within Brazilian wind-energy contexts.

Newcastle University (UK) contributes specialist expertise in electromagnetic NDT and thermographic data processing, supporting advanced excitation concepts and robust reconstruction approaches for complex geometries. Newcastle also supports dissemination activities, including joint seminars and workshops on thermal NDT and AI-enabled data interpretation.

Brazil is expanding wind energy as part of its long-term decarbonisation pathway, and wind turbine blades (WTBs) are critical assets whose integrity directly affects safety and operational efficiency. Thermography is a widely used non-destructive evaluation (NDE) technique for WTB inspection: a controlled heat stimulus is applied and the resulting surface temperature evolution is monitored to infer subsurface anomalies that alter thermal transport behaviour ^[1,2].

In real blades, conventional thermographic inspection faces key challenges. First, complex curved geometry (especially corners and edges) can cause non-uniform heating and shadowing effects, reducing the reliability of defect characterisation ^[1,2]. Second, accurate interpretation requires robust forward heat-transfer modelling, which becomes computationally demanding and error-prone in complex geometries and for anisotropic composite materials ^[3]. Third, many inversion approaches rely on simplified heat-transfer assumptions; their accuracy can degrade when heat flow is influenced by complex geometries, heterogeneous materials, and varying boundary conditions ^[4].

Figure 1 provides a high-level view of the project context and the overall inspection-to-interpretation workflow for curved composite blades, motivating the need for geometry-aware thermography reconstruction and inversion.

To address the above challenges, the project aims to improve the effectiveness of wind turbine blade inspections by developing next-generation thermography reconstruction and inversion methods, leveraging complementary UK and Brazilian expertise. The focus is on improving the reliability of thermal data acquisition on curved composite surfaces, and enabling more interpretable, physics-consistent reconstruction for defect assessment ^[3,4].

As illustrated in Figure 2, the methodology includes enhancing thermal excitation and measurement quality for curved blades. This is achieved through advanced excitation strategies, such as coded/pulse-compression concepts that improve signal-to-noise ratio and resolution, and coil/excitation configurations designed to better adapt to curved geometries ^[5]. These developments target more uniform, repeatable heating and more informative thermal responses for subsequent interpretation.

Figure 3 summarises the reconstruction and inversion direction: the project advances thermography reconstruction by combining physics-based understanding with modern learning-based tools, aiming for robust performance under realistic heat-transfer conditions. In particular, wave-inspired thermal reconstruction concepts and physics-informed learning (for example, incorporating physical loss and constraints during model training) are used to improve the fidelity and interpretability of reconstructed results while remaining adaptable to different blade designs and operating scenarios ^[6].

This ICA Enhancement Funding project extends the existing ISPF UK–Brazil programme on thermographic reconstruction for wind turbine blade inspection. The enhancement adds two elements that are essential for translation but are not fully resourced in the core award: independent ground-truth validation and an industrial deployment and trial package in Brazil.

Northumbria University (UK) will lead the enhanced thermography reconstruction and virtual aperture focusing work, reformulating reconstructed virtual thermal wavefields as a thermal analogue of full matrix capture (FMC) and developing a curvature-consistent total focusing method (TFM) analogue for defect imaging.

University of Bristol (UK) will provide independent ultrasound FMC datasets and TFM reconstructions for selected representative cases, creating a separate ground-truth route for cross-modality validation. Prof. Paul Wilcox will provide senior academic mentoring and methodological guidance for composite NDT, imaging, inverse problems, and validation protocol design.

Federal University of Rio de Janeiro (Brazil), together with WEFA ENGENHARIA, will support thermography deployment, calibration, traceability, and factory-based trials, linking the enhanced methodology to repeatable industrial inspection practice in Brazil.

The existing programme reconstructs virtual thermal wavefields from thermography on curved wind turbine blade geometries using regularisation and learning-based methods. The enhancement addresses the next stage of the technical pathway: moving from reconstructed wavefields toward a robust imaging and validation framework that can provide stronger evidence of defect location, geometry, accuracy, and uncertainty.

The reconstructed virtual thermal wavefields will be organised as a thermal analogue of full matrix capture (FMC), followed by a curvature-consistent total focusing method (TFM) analogue for virtual aperture focusing and defect imaging. Physics-attention time-of-flight control, locked calibration settings, and point-spread-function-based calibration will be used to stabilise focusing, suppress diffusion artefacts, and support reproducible robustness and uncertainty assessment.

Independent cross-modality validation will use Bristol-hosted ultrasound FMC datasets and TFM reconstructions for selected representative cases. Matched cases will be co-registered so that thermography-derived results can be benchmarked against an independent reference using defect localisation, geometry descriptors, cross-modality consistency, and failure-mode analysis.

The validation framework will explicitly account for curved blade-like geometries, while the Brazil workstream will develop repeatable thermography deployment procedures for industrial trials, including calibration, emissivity control, safe excitation and capture, and traceable data logging. UK–Brazil researcher mobility and training will support capability transfer across the partner institutions. Planned outputs include benchmark datasets, reproducible analysis scripts, a validated virtual aperture focusing workflow, and an industry-usable thermography inspection procedure and trial report.

1. G.Y. Tian., et al. IEEE Transactions on Industrial Electronics 63.10 (2016): 6305-6315.
2. Cesar Giron Camerini et al. Journal of Nondestructive Evaluation 40.3 (2021): 58.
3. Yi, Q., et al. Composites Part B: Engineering 178 (2019): 107461.
4. Yi, Q., et al. NDT & E International 122 (2021): 102474.
5. Yi, Q., et al. NDT & E International 102 (2019): 264-273.
6. Thummerer, G., et al. NDT & E International 112 (2020): 102239.