City Research Online

Abstract

This thesis develops a physics-informed structural health monitoring (SHM) framework for aerospace composite structures that progresses from controlled damage characterisation at the coupon scale to operationally realistic assessment under authentic flight loading. Carbon Fibre Reinforced Polymer (CFRP) composites offer substantial weight and performance benefits for aerospace applications, but their susceptibility to complex, interacting damage mechanisms—such as delamination and ply-level material degradation—poses significant challenges for reliable in-service inspection and monitoring.

Much of the existing SHM literature relies on idealised loading conditions and isolated damage modes, limiting its applicability to real aircraft structures operating under time-varying aerodynamic, inertial, and environmental loads. This work addresses these limitations through a three-stage investigation that systematically increases modelling fidelity, damage complexity, and operational realism.

The first stage establishes fundamental delamination characterisation using a meso-scale 17-layer quasi-isotropic CFRP coupon (1 m × 1 m × 10 mm) instrumented with surface-mounted piezoelectric sensors. Three-dimensional finite element simulations generate 5,865 synthetic damage cases spanning variations in delamination location, size, and through-thickness position. Derivative-based feature engineering using Gaussian kernel smoothing is combined with deep neural networks to achieve accurate spatial localisation, with mean absolute errors below 0.032 m, and reliable estimation of damage severity under rigorously stratified data splits.

The second stage extends the framework to the more realistic and challenging scenario of concurrent damage, in which delamination and ply-level material degradation coexist within the same structural region. Over 105 parametric simulations are generated by simultaneously varying delamination geometry and local stiffness reductions used as a proxy for material damage. Rather than treating damage mechanisms independently, all models are trained and evaluated exclusively in the both-present regime. A task-decomposed learning strategy, supported by gradient-augmented sensor features, enables robust single-snapshot inference of interacting damage mechanisms, achieving coefficients of determination of approximately R² ≈ 0.79 for damage localisation and R² ≈ 0.87 for damage size estimation using only five ultrasonic sensors.

The final stage integrates operationally realistic loading derived from a complete Boeing 777–200 transatlantic flight profile. Time-varying aerodynamic pressures are reconstructed using angle-of-attack polar inversion, validated pressure coefficients, progressive fuel-mass evolution, and engine-induced dynamic excitation, and applied to a validated NASA Common Research Model wing-box with composite skins. Three complementary sensing and analysis approaches are investigated: (i) multiview image-based deformation analysis for global damage screening, (ii) sparse global piezoelectric sensor networks for moderate-resolution localisation, and (iii) global-to-local coupling that transfers wing-level structural response to high-resolution mesoscale panels for detailed damage characterisation.

Across all scales, the results demonstrate that physically informed feature design and explicit modelling of damage interaction are critical for reliable SHM under realistic loading conditions. The work establishes a unified framework capable of bridging laboratory-scale studies and operationally representative simulations, providing a robust foundation for future development of scalable, data-driven SHM systems for composite aircraft structures.

Publication Type:	Thesis (Doctoral)
Subjects:	Q Science > QA Mathematics > QA75 Electronic computers. Computer science T Technology > TA Engineering (General). Civil engineering (General) T Technology > TL Motor vehicles. Aeronautics. Astronautics
Departments:	School of Science & Technology > Department of Engineering School of Science & Technology > School of Science & Technology Doctoral Theses Doctoral Theses

Export

Downloads