This thesis investigates the mitigation of in-plane fiber waviness in the press consolidation of Carbon Fiber Polyetherketoneketone (PEKK) V-tail front spars. The manufacturing process includes Automated Tape Laying (ATL), preforming, and press consolidation. The study identifies that fiber waviness originates during the press consolidation phase, particularly on the outer mold line (OML) side. Three hypotheses were investigated: waviness due to matrix flow, flange compression, and tool-part interaction during tool expansion. Mitigation strategies include installing guidance pins, placing shim material, modifying the pressurization sequence, central fixation of the female tool, and spot welding the ATL stack. These actions successfully eliminated fiber waviness, as confirmed by subsequent consolidations. An image processing algorithm was developed for consistent measurement. The study concludes that uniform compaction, uniform tool gap, and maintaining a consistent temperature range are essential to preventing fiber waviness in composite spar manufacturing.

Injection molding is a widely used process for manufacturing a large range of polymer parts, including fiber-reinforced polymer composites. The complex nature of this process, combined with the non-Newtonian behavior of molten polymers, implies significant challenges in terms of control, prediction, and optimization of the processing parameters that influence manufacturing quality. While various computational and numerical models have been proposed to simulate the injection molding process and ensure the desired part quality (see, for instance, industrial software such as Moldex3D or Moldflow, they bring their own challenges, including high computational costs, the need for solver tuning, frequent parametric studies, and the economic burden associated with licensing.

In this study, machine learning surrogates are explored as an alternative to conventional modeling software. They are based on Graph Neural Networks (GNNs) that have recently demonstrated remarkable potential for simulating physical systems. In a graph, physical systems are encoded as nodes end edges, capturing their organization. Then, state-of-the-art techniques such as the message passing framework are tested for their potential to learn physics. In the long term, this project aims to establish a novel graph-based framework capable of efficiently simulating complex physics across various material models and processing conditions.

A numerical model is developed for 2D mold filling and quantitively validated against a benchmark. The prediction results demonstrate that the learned simulator is able to capture the physics of melt front behavior. Further, the prediction runtime is significantly faster than the ground truth, completing in under 1 second compared to \textasciitilde hours, representing a substantial improvement in efficiency. However, the accuracy of the predictions can vary, with errors accumulating especially during testing. While the scalar loss metric provides an estimate of the model's performance, it lacks expressiveness and can be unreliable on its own.

Heterogeneous materials are vital for both the modern engineer and inquisitive scientist alike. They make up a vital material class that can either form inevitably as a result of material processing (such as crystallization in metals) or can be intentionally designed for to gain desirable properties (such as anisotropy in composites). As such, due to their prevalence and applicability to various engineering problems, predicting their behavior has become a topic of great interest in the solid mechanics community over the last few decades.

One particularly pressing challenge is performing fast mechanical simulations along multiple length scales in heterogeneous materials. To this end, the reduced order method known as Self-Consistent Clustering Analysis (SCA) has been proposed as an effective means of striking a balance between accuracy and efficiency. Underpinning this is SCA's ability to decompose a domain into clusters in a preliminary offline stage (learning), efficiently reducing the problem's degrees of freedom. It has been shown to be remarkably accurate in predicting plasticity without significantly degrading accuracy compared to other methods, such as the finite element method. However, the offline stage requires a quantity known as the Cluster Interaction Tensor (CIT) to be computed, whose computational complexity scales quadratically with the number of clusters, thus causing a bottleneck in the method. Additionally, the CIT is recomputed during the online stage (prediction) in the recently proposed extension called adaptive Self-consistent Clustering Analysis (ASCA), which further stresses the need to speed up their computation.

To address this limitation, a data-driven surrogate model is proposed to compute the CIT efficiently. The behavior of the local CIT components is first analyzed, from which it is concluded that a surrogate model shall be developed to predict upper off-diagonal terms. This decision is made due to the bi-modal nature of certain components in the tensor and the quadratic scaling behavior of off-diagonal terms versus the linear scaling of diagonal ones. Following that, a sensitivity study shows how various magnitudes of noise affect the homogenized response's accuracy and convergence performance. Furthermore, it is shown that the solution accuracy and convergence behavior are degraded when the number of clusters is increased. With a proper understanding of the CIT's behavior and its function within SCA, the surrogate can be created.

The surrogate's feasibility is first shown using the ResNet-18 architecture, a CNN model derived from computer vision. It is demonstrated that ResNet-18 can make accurate predictions for a range of different microstructural parameters. This includes the number of clusters and samples in the dataset's distribution and out of distribution. Composite RVEs are used as out-of-distribution samples to test the robustness of the surrogate to realistic inputs. An attempt is made to improve the model's performance by training it on an unbalanced dataset, and although it improves the overall CIT prediction in specific regimes, the resulting stress-strain generalizability is degraded. Subsequently, a lighter version of ResNet-18, known as ResNet-lite, is tested and shown to give faster predictions than the baseline method. This, however, comes at a cost to the accuracy of the solution.

Additionally, deemed as a critical aspect of the study, the efficient generation of a large representative training dataset is discussed. A novel method for generating clustered microstructures efficiently using gradient noise is introduced. By leveraging datasets derived using this method, the surrogate can learn the fundamental interactions needed to make accurate predictions on realistic, out-of-distribution samples. The need for a large dataset is further emphasized using a dataset size sensitivity study.

Stiff and transparent materials are essential across industries such as aerospace, defense, and consumer electronics. Traditional materials like glass and ceramics, while effective, have limitations due to brittleness, high density, and low impact resistance. Consequently, there is a demand for lightweight, durable materials with favorable optical properties. Fiber-reinforced polymers (FRPs) offer high strength-to-weight ratios and greater impact resistance than glass and ceramics. Given the low light absorption of glass fibers in the visible spectrum, transparent glass fiber-reinforced polymers (TGFRPs) show promise as replacements for traditional transparent materials.

This thesis explores two main aspects: (1) optimizing TGFRP transparency through manufacturing parameters and (2) leveraging TGFRP transparency for damage detection and stress analysis. In the first phase, findings demonstrate that achieving high surface smoothness and maximizing light transmission within the green spectrum (520–600 nm) are crucial for enhancing TGFRP transparency. This is best controlled via post-curing, a more stable approach than adding methyl methacrylate (MMA) as suggested in existing literature. Sizing was also found to play a critical role in transparency, ensuring optimal bonding between glass fibers and the matrix. Additionally, minimizing the number of glass fiber fabric layers improved transparency by reducing transmission losses across visible wavelengths.

The second phase investigated the transparency of TGFRP for visualizing internal damage and stress distributions. Confocal microscopy was used to observe cracks at various depths within TGFRP samples, while image processing techniques, such as thresholding, were employed to calculate crack density. The results indicate that a higher crack density corresponds to reduced light transmittance across the visible range. However, some limitations were encountered, as cracks and scratches on the upper layers cast shadows on the lower layers, complicating the detection of deeper damage.

During tensile testing of open-hole specimens, significant opacity changes were observed in high-strain regions, suggesting that opacity correlates with localized stress. Digital image correlation (DIC) and grayscale histogram analysis provided insights into the strain thresholds at which opacity changes begin. Future work could build on this by examining the extent to which opacity changes are due to refractive index variations alone, independent of crack formation, using these strain levels as a baseline.

Furthermore, verifying the reversibility of opacity changes through post-test spectrophotometry could position TGFRPs as highly effective materials for developing innovative and reliable non-destructive testing, damage detection, and stress visualization methods in glass fiber-reinforced polymer composites.

This master’s thesis focuses on the design and assembly of an innovative tow spreading line for investigating carbon fiber tow spreading through the utilization of spreader bars. The primary objective was to examine the friction behavior during the spreading process by employing multiple tension sensors. Additionally, novel monitoring concepts including 4-point resistivity sensing, optical width and gap detection, and optical analysis for determining fiber orientation within the carbon tow were introduced. Through a series of experiments conducted on the newly built setup, comprehensive data was collected and analyzed. The findings revealed an intriguing observation that deviates from the established Capstan equation. It was observed that an increase in tension in the tow resulted in a reduced apparent friction coefficient during bar tow spreading. This departure from the conventional understanding of friction dynamics in this scenario contributes valuable insights to the field of carbon fiber tow spreading. Furthermore, the feasibility of utilizing resistivity measurements as a means of detecting material anomalies, such as damage or waviness, was investigated. The results demonstrated that this technique, while partially reliant on tow tension, consistently detected such anomalies. However, challenges were encountered in achieving quantitative consistency in the optical orientation analysis, making it difficult to obtain robust results in this aspect. The experimental monitoring setup also revealed an increase of approximately 20% in the width of the tow. To enhance process repeatability, recommendations are proposed to upgrade several components of the built experimental line. Moreover, it is advised to further test and develop the software analysis techniques to achieve a higher degree of repeatability, thus potentially validating the obtained results. In conclusion, this research contributes to the understanding of carbon fiber tow spreading through the design and assembly of an experimental production line, examination of friction behavior, and the introduction of novel monitoring techniques. The outcomes serve as a foundation for future investigations and advancements in this domain, with the potential to enhance the efficiency and quality of carbon fiber tow spreading processes.

Frontal polymerization (FP) has emerged as a promising alternative to traditional bulk curing methods in recent years for manufacturing high-performance fibre-reinforced polymer (FRP) composites. The energy utilized in this self-propagating curing strategy is solely derived from the exothermic enthalpy of polymerization, making it potentially more efficient than traditional curing methods, which are extremely energy-intensive and therefore unsustainable. Research in the field of FP has been primarily focused on studying the effectiveness of the FP formulations, particularly the relationship between the monomer types and initiator concentrations on front properties. However, the research on the use of FP for manufacturing FRPs has been limited to flat rectangular plates. Therefore, this research aims to investigate the behaviour of the propagating fronts in composites with varying fibre volume fractions (Vf) within the sample and varying geometries to assess the feasibility of using FP in structures that more closely resemble real-life applications. Several test setups were explored to determine the best method for maintaining the heat balance required to sustain a propagating front. Through these trials, a stainless steel-based closed mould test setup with Teflon sheets was conceptualized and manufactured, consistently allowing for the manufacturing of composite samples with a Vf below 36%. The influence of changing Vfs on the behaviour of the front was studied, and the results were quantified using temperature data captured via thermocouples placed at strategic locations. Through a series of experiments, a critical length was established that allowed the propagation of the front from the low Vf region to the high Vf region. When the length of the low Vf region with respect to the high Vf region was smaller than the critical length, the front was seen to quench at the interface. This phenomenon was attributed to the physical reduction in the volume of the resin, which implied a reduction in the generated heat through polymerization. This led to fronts with lower peak temperatures, which upon reaching the interface would quickly fall below the threshold temperature required to sustain the front due to the higher rate of heat loss resulting from the higher fibre content. This study was followed by the manufacturing of L-shaped composite samples with uniform Vf, which showed the propagation of the front in complex geometry. Finally, the Vf was varied within the L-shaped samples, and the behaviour of the front showed similarities to the rectangular samples with varied Vf, leading to the conclusion of the presence of a critical length irrespective of geometry. These findings effectively contribute to the knowledge base necessary for implementing FP as a curing strategy for FRPs. The results obtained from this study can inform the design of manufacturing processes and setups tailored for the use of FP, potentially leading to more efficient and sustainable manufacturing practices.

In recent years, lightweight composites have become increasingly popular in several industries for the benefits they offer such as high strength to weight ratio, cost efficiency, tailorable properties and superior mechanical performance. The performance of these lightweight composites can be further optimized by using spread tows as a reinforcement solution. Tow spreading is a process of flattening fibre tows into thinner and wider tapes with a low aerial weight which increases resin impregnation of fibres, lowers void content in impregnated tows and produces lightweight composites with improved damage tolerance.

The majority of tow spreading technology however is owned by companies in the form of intellectual property and patents thus limiting the scientific understanding of the process. This aspect leaves tow spreading open as a field of research for exploration and investigation. While some previous studies have investigated the process of tow spreading, the applicability of these studies has so far been limited in their experimental apparatus. These studies identified the gap that exists between the sophisticated industrial technology and the research setups hence providing an incentive to further develop an experimental facility that generates a deeper understanding of the current industrial practices in tow spreading.

In this research, a mechanical tow spreader was developed that utilizes cylindrical spreader bars to facilitate the spreading of fibres. This was followed by an experimental analysis where different input parameters in the tow spreader such as tension, spreader bar orientation, heat and speed among others were investigated to understand their influence on spreading behaviour. Lastly, a microstructural investigation of spread tow was conducted where a technique to capture the spread of fibres in the tow was developed and microstructural samples were generated. These samples were then observed under a microscope to visualize the evolution of fibre distribution as the tow spreads.