Thesis
Rework of gap and overlap defects in automated fibre placement manufacturing
- Creator
- Rights statement
- Awarding institution
- University of Strathclyde
- Date of award
- 2026
- Thesis identifier
- T18107
- Person Identifier (Local)
- 202291341
- Qualification Level
- Qualification Name
- Department, School or Faculty
- Abstract
- Automated Fibre Placement (AFP) manufacturing is the current state-of-the-art for high-rate manufacturing of high-performance composite parts. The method relies on the robustness and repeatability of robotic elements to enable rapid and reliable manufacturing of composite parts for a range of industries, including aerospace, automotive, renewables, space, defence and marine. While improvements in deposition rates are achieved through AFP compared to manual part manufacturing, the technique remains challenged by the presence of defects in manufactured parts, many of which rely on manually performed inspection and critically, rework. Importantly, the inspection and rework of these defects accounts for a disproportionately high amount of the cycle time. While the hallmark of AFP is its level of automation, surpassing most other composite manufacturing techniques, the process for inspecting and addressing these defects remains largely manually performed by skilled operators. This manual task necessarily requires the shutting down of automated production to ensure the safety of the operators, contributing to the reduction in manufacturing efficiency incurred by inspection and rework. Various research efforts have been performed for the development of automated defect inspection methods; little research, however, has been undertaken on the subsequent reworking step. Additionally, and critically, the effect of this reworking on part quality is not well understood and has not been rigorously studied prior to this Thesis. This Thesis addresses this gap, firstly through an analysis of the process and effect on part quality, both in structural and mechanical terms, of rework; and further to develop innovative automated methods for performing the reworking activity itself. Through this work, the fundamental research question of the effect of reworking defects under current best practices is answered, while paths towards automated reworking are developed and tested. In current practice, manual rework is performed by skilled human operators through a variety of techniques. The techniques themselves are poorly understood in the literature and instead exist as “crafts” of sorts; understood implicitly by the human practitioners themselves, but not documented, nor optimised. This research includes the first substantial body of research into this activity, to understand the key features and optimise the activity. This Thesis provides the first description of the existing knowledge both in the literature and through industry perspectives on the rework activity and provides a first-of-its-kind rework process analysis, resulting in a set of best practices for performing the activity while minimising the damage to the surrounding and removed material. Secondly, through various non-destructive and destructive testing procedures the effect of rework is examined. With the exploratory literature and experimental research on the manual reworking activity conducted, a series of test cases were developed in order to evaluate the effect of reworking on a composite part. These test cases were analysed through different quality analysis methods. Firstly, ultrasonic non-destructive testing (NDT) was used to demonstrate the variance in out-of-plane waviness amongst reworked specimens and also compared to pristine and defective specimens. The reworked specimens, while improved in terms of out-of-plane waviness from the defective benchmark, still fell below the quality of pristine specimens, highlighting that reworking is both variable and not necessarily a direct guarantee of pristine-like quality. Further analysis through the ISO 14125 flexural tests revealed that reworking effectively restored much of the stiffness of the laminates to near-pristine levels, but a higher level of variance in results was seen, mirroring the non-destructive test results. Further evaluation of the effectiveness of rework was performed by, firstly, analysing the microstructure of pristine, defective and reworked laminates through Scanning Electron Microscopy (SEM). The results showed that the reworked laminates displayed a higher degree of inter-ply disturbances centred around the rework sites, despite featuring reduced local variations in thickness compared to defective specimens. This was particularly the case where rework was performed on subsequent piles with identical fibre angles. A novel Benchtop-AFP system was developed as a research platform to enable lab-scale laminate manufacture and was used for the Ultrasonic NDT test laminates. This collaborative robot-based system provided the flexibility and accessibility required for iterative experimentation, allowing for the codification and optimisation of rework strategies within a semi-automated environment. Beyond its application in this research, the Benchtop-AFP concept represents a contribution towards democratising AFP research, offering a scalable model for laboratory-scale studies of automated composite manufacturing. Finally, from the literature review and expert outreach, evidence of significant space for the development of automated reworking methods exists. Two methods are investigated within this Thesis. Firstly, a recompaction based gap defect rework method is explored in terms of the effect of varying processing parameter levels through statistical analysis. The analysis showed an unintended coupling between the width change deformation and undesirable transverse displacement of the tow, while the overall magnitude of width change was insufficient for effective gap defect mitigation. Secondly, a real-time trajectory control method was developed making use of an expanded motion control toolkit and a ROS-hosted end-to-end defect detection to correction system. Using this method, effective mitigation of the extent of gap defects was achieved during the initial deposition step, within the constraints of the system design. Within the scope of this thesis is the analysis of the rework activity specifically for defects occurring during automated fibre placement. These are limited to placement and process induced defects and excludes material-based defects. This Thesis demonstrates that while manual rework offers partial remediation of AFP-induced defects, the results of the process are highly variable. By codifying manual rework practices and systematically evaluating their impact on part quality, this work provides a foundation for standardised study and optimisation. Furthermore, it pioneers automated rework methods, including tow reconsolidation techniques and a real-time trajectory correction framework based on sensor-informed control, marking a significant step toward fully autonomous AFP. These contributions collectively address a critical gap in AFP manufacturing and advance its readiness for industrial deployment in high-performance composite applications.
- Advisor / supervisor
- Yokan, Catherine
- Mehnen, Jörn
- Resource Type
- DOI
Relations
Items
| Thumbnail | Title | Date Uploaded | Visibility | Actions |
|---|---|---|---|---|
|
|
PDF of thesis T18107 | 2026-08-05 | Public | Download |