uBibliorum
Repositório Digital da Universidade da Beira Interior
Entradas recentes
Graphene-based composite laminates for structural applications
Publication . Parente, João Miguel Nunes; Reis, Paulo Nobre Balbis dos; Silva, Abílio Manuel Pereira da; Pereira, João Pedro Nunes
This thesis explores the synthesis, microstructure and properties of polymer-matrix composites reinforced with graphene nanoplatelets (GNP), carbon nanotubes (CNT) and carbon nanofibers (CNF), both as individual fillers and in hybrid combinations. The aim is to develop structural materials with high mechanical performance and integrated self-sensing capabilities. Nanocomposites were produced by dispersing nanoparticles in epoxy resin, followed by thermal curing. Mechanical and sensing behaviour were evaluated using three-point bending and cyclic fatigue tests under realistic loading conditions.
Viscosity measurements show that GNP or CNF can increase resin viscosity by up to 74 %, which challenges processing at higher loadings. Curing at 5 °C yields the best flexural strength and surface hardness, with GNP-reinforced composites showing up to 24.8 % strength improvement. Volumetric shrinkage during cure varies with filler type: GNP increases shrinkage by 91 %, while CNF reduces it by 77 %, underscoring the role of nanoparticle–matrix interactions.
In fibre-reinforced laminates, placing glass fibres on the compression side of hybrid carbon/glass layups enhances both load and deflection at peak, with a 3G/5C configuration achieving a 5.9 % higher peak force and 13.1 % greater deflection than the reverse. Incorporating 0.75 wt. % GNP into carbon-fibre laminates raises bending stiffness and extends fatigue life by 15.1 %, while increases of 10.6 % and 9.2 % are observed in hybrid and glass-fibre laminates. These enhancements show that even low graphene concentrations effectively delay fatigue damage.
Combining CNT and GNP at equal loadings (0.375 wt. % each) yields synergistic gains in stiffness and strength. In carbon-fibre composites, strength and stiffness rise by 11 % and 18 %, respectively; glass-fibre composites show gains of 8 % and 55 %. Electrical resistance monitoring during fatigue reveals gauge factors up to three times those of commercial strain gauges, with stress-amplitude sensitivity improving by 20.1 % in carbon composites and 32.4 % in glass composites. This confirms high-sensitivity and stable self-sensing.
The integration of hybrid nanofillers with mechanical and electrical performance validation forms a framework for designing multifunctional composites. Key challenges remain in scaling uniform nanoparticle dispersion, ensuring long-term network stability and standardizing sensor calibration. Addressing these issues is essential for translating laboratory results into practical Structural Health Monitoring (SHM) applications.
Desenvolvimento, caracterização e otimização de um biocompósito cerâmico multifuncional
Publication . Macedo, Duarte Félix; Silva, Abílio Manuel Pereira da; Oliveira, Filipe José Alves de; Oliveira, Mariana Braga de
Esta tese foca-se no desenvolvimento e otimização de biocerâmicas de Fosfato Tricálcico (TCP) dopado com iões metálicos e reforçado com grafeno ou zircónia, com o objetivo de criar materiais altamente biomiméticos para aplicações em engenharia de tecidos ósseos. O TCP foi dopado com iões de Mg2+, tendo obtido melhores propriedades mecânicas para TCP + 10 mol% Mg2+. Seguidamente, foi dopado com iões Mn2+, Zn2+ e Fe3+, cujas combinações foram otimizadas através do Design of Experiments (DoE) para combinação conjunta de 5 mol% total de dopagem, revelando que certas proporções promovem microestruturas densas, com boas propriedades mecânicas e ausência de citotoxicidade. Das várias composições testadas destacou-se o TCP dopado com 10 mol% Mg2+, 1,67 mol% Mn2+, 1,67 mol% Zn2+ e 1,67 mol% Fe3+ (TCP7). A adição de zircónia tetragonal e cúbica aumentou a resistência mecânica em até 55%, enquanto a porosidade interconectada, controlada com o porogénio Polimetilmetacrilato (PMMA), simulou a arquitetura óssea natural, favorecendo a viabilidade celular, mas reduziu a resistência mecânica em mais de 90%. Paralelamente, a incorporação de grafeno, até 3 wt%, melhorou significativamente a resistência à compressão diametral sem comprometer a biocompatibilidade, demonstrada em ensaios de citotoxicidade. A composição com melhores propriedades mecânicas foi o TCP7 com 1,5 wt% de grafeno. Para apoiar o design racional destes materiais, foi desenvolvida uma abordagem de modelação numérica avançada baseada em microestruturas reais 2D, permitindo gerar Volume Elementar Representativo (RVE) em 3D que preveem com alta precisão o módulo de Young dos compósitos, superando métodos analíticos convencionais. Em conjunto, os resultados demonstraram que é possível conceber biocerâmicas de TCP com propriedades mecânicas reforçadas, arquitetura porosa controlada e elevada citocompatibilidade, graças a uma estratégia multifatorial que combina dopagem iónica, reforço com nanomateriais e simulação digital. Estes avanços posicionaram os materiais desenvolvidos como candidatos promissores para implantes ósseos multifuncionais, com forte potencial clínico em ortopedia e odontologia.
Use of the microalgae-bacteria Consortium in photobioreactors for the treatment of wastewater from the paper pulp industry
Publication . Sátiro, Josivaldo Rodrigues; Albuquerque, António João Carvalho de; Simões, Rogério Manuel dos Santos
This thesis investigated the application of microalgae–bacteria consortia in photosynthetic and granular systems for the treatment of pulp and paper industry wastewater, characterized by low biodegradability and reduced nitrogen content. The study evaluated how inoculum concentration affects start-up, hydraulic retention time (HRT), and initial reactor sizing. Optimal microalgae-to-bacteria ratios (1:1, 1:5, 3:1) were identified: PBR5 (1:5) promoted higher COD and nitrogen removal, PBR6 (3:1) enhanced phosphorus removal, and PBR3 (1:1) improved biomass formation. Excessive microalgae hindered granule formation, while an optimized HRT of 16 h with 8 h cycles was determined. In photobioreactors, higher bacterial proportions accelerated organic matter degradation (up to 85% COD removal) and promoted efficient flocculation (>90%), whereas higher microalgae ratios improved phosphorus removal (up to 86%) and lipid accumulation in the biomass (22%). The symbiotic interaction also enabled nitrogen removal above 85% via nitrification and assimilation. A hybrid system combining photobioreactors and constructed wetlands treated raw industrial effluent, achieving 89% COD, 69% total nitrogen, 59% total phosphorus, and 81% phenolic compound removal. The system maintained high operational stability (600 mgVSS/L biomass) and excellent settleability (89.7%), demonstrating the potential of nature-based solutions and integration into circular economy strategies. Aerobic granular sludge systems with microalgae–bacteria consortia (AB-AGS) were evaluated in sequential reactors, showing rapid granulation (>82% of particles >1.0 mm in 15 days), high compactness, and structural stability, with phosphorus removal above 92% and significant organic matter and nitrogen elimination. Operational comparisons highlighted differences in settleability: one system reduced the sludge volume index (SVI) from 31.84 to 4.59 mL/g, while another maintained 27.42 mL/g, indicating slower but functional compaction. The latter also exhibited higher lipid accumulation (23.3 ± 1.8%), more than double that observed in reactors inoculated solely with bacteria (9.3 ± 1.7%), demonstrating the potential of AB-AGS as an integrated biorefinery for biofuel precursors without chemical additives. Overall, the results confirm that integrating microalgae–bacteria consortia into photosynthetic, hybrid, and granular systems is an effective and sustainable approach for treating complex industrial wastewater. These systems enable high pollutant removal, generate valuable biomass, and reduce reliance on energy-intensive processes. Although conducted at laboratory scale, the findings provide a solid foundation for future pilot-scale validation and potential industrial implementation.
Self-Explanatory Deep Learning Models with Concept-based Multimodal Explanations for Medical Imaging Diagnosis
Publication . Pires Patrício, Cristiano; Neves, João Carlos Raposo; Teixeira, Luís Filipe Pinto de Almeida
The remarkable performance of deep learning models in automated medical imaging diagnosis is achieved at the expense of the low interpretability of their representations. The opaque nature of these methods, which often operate as “black boxes”, remains a major barrier to their adoption in real-world applications, especially in high-stakes scenarios such as healthcare. This lack of interpretability motivated the development of eXplainable Artificial Intelligence (XAI) techniques capable of explaining model decisions so that humans can understand and interpret their decision-making. Early efforts in XAI applied to images relied mainly on post-hoc strategies that generate model-agnostic explanations by assessing the influence of input regions on predictions. However, these explanations are often ambiguous and unreliable. Similarly, textual explanations face challenges as language models are prone to generate inaccurate content, including ambiguous or factually incorrect statements. As an alternative, Concept Bottleneck Models (CBMs) offer an inherently interpretable design, where the final predictions are explicitly derived from intermediate human-understandable concepts. Nevertheless, CBMs face several critical limitations. Their reliance on manual concept annotations, the lack of visual interpretability for the predicted concepts, and the need for model retraining when new concepts are introduced hinders their utility and scalability. This thesis addresses these limitations by introducing methods capable of generating multimodal explanations grounded on human-understandable concepts, thereby enhancing both the transparency and the interpretability of the model output. First, we present a comprehensive survey of state-of-the-art XAI methods, datasets, and evaluation metrics in medical image diagnosis, highlighting existing gaps and open challenges in the XAI literature. Building on these insights, we propose two concept-based approaches for skin lesion diagnosis: one extending the conventional CBMs to produce concept-based visual explanations, and another that leverages a transformer-based architecture with learnable concept tokens, improving the visual coherence of concept explanations through a dedicated architecture and regularization. To reduce reliance on concept annotations, we further explore Vision Language Models (VLMs), proposing strategies that automatically annotate concepts and predict the final diagnosis either through a linear classifier or by prompting Large Language Models (LLMs). To overcome the lack of visual context in disease prediction in these latter approaches, we propose CBVLM, a training-free framework that integrates off-the-shelf Large Vision-Language Models (LVLMs) to jointly generate concept-based explanations and predict disease diagnoses grounded in both semantic concepts and visual demonstration examples. Beyond concept-based explanations, we also demonstrate that interpretability can also be achieved even in constrained scenarios with limited annotations. Specifically, we propose an unsupervised framework for brain Magnetic Resonance Imaging (MRI) tumor detection that learns to reconstruct benign patterns of an input image using solely a dataset of healthy examples. At inference, when presented with brain MRI containing anomaly patterns, the reconstruction error between the input and the reconstructed image highlights potential tumor regions, allowing intuitive and interpretable anomaly localization. The results obtained from the methods proposed in this thesis demonstrate that it is possible to enhance the interpretability of CBMs by integrating visual concept explanations consistent with the learned concepts, while reducing their reliance on manual concept annotations, maintaining the interpretability and performance. Furthermore, extensive experiments across various medical imaging modalities, including dermoscopy, radiology, eye fundus imaging, and brain MRI, demonstrate that the proposed approaches not only improve disease diagnosis, but also provide more transparent and faithful multimodal explanations, paving the way for safer clinical integration and increased trust.
House dust mite molecular sensitisation profile and allergic respiratory disease expression
Publication . Semedo, Filipa Alexandra de Matos Tavares; Inácio, Filipe Fernando da Cruz; Barata, Luís Manuel Taborda
Background The molecular era of allergology has transformed our understanding of IgE sensitisation, enabling precise identification of allergen components through component-resolved diagnostics (CRD). House dust mite (HDM) and storage mite (SM) allergens are major drivers of allergic asthma (AA) and allergic rhinitis (AR), yet their individual contributions to clinical expression and evolution of disease remain incompletely understood.
Objectives This thesis aimed to characterise the clinical relevance of individual mite molecular components in allergic respiratory diseases by combining high-level evidence synthesis, in vivo functional assessment, and real-world cross-sectional and longitudinal analysis.
Methods A systematic review and meta-analysis (SR/MA) of 101 studies (n = 23 781) was conducted to assess the relationship between sensitisation to major HDM and SM components and clinical phenotypes. The clinical impact of newly identified major allergen Der p 23 was assessed trough monosensitisation to this allergen in allergic respiratory patients with no other sensitisations. Functional relevance of SM Lepidoglyphus destructor (Lep d) was investigated using nasal provocation testing (NPT) with Lep d in urban, non-occupational patients. The clinical impact of Lep d was also analised through monosensitisation to Lep d 2 in allergic respiratory patients. Lastly, a 20-year longitudinal cohort study compared the evolution of mite molecular sensitisation profiles and respiratory symptom dynamics in adult population and comparison with children and adolescent cohorts.
Results The SR/MA revealed that major mite allergens as Der p 1, Der p 2, Der p 23, Der f 1, Der f 2, Lep d 2 and Blo t 5 have higher sensitisation rates and sIgE concentrations in patients with AA than in those with only AR. In vivo testing with NPT and Lep d 2 monosensitisation confirmed that Lep d is clinically relevant in urban settings in patients with allergic respiratory disease. Monosensitisation to Der p 23 showed potential clinical impact, highlighting overlooked profiles in routine diagnosis. The longitudinal study demonstrated molecular involution in adults—marked by declining sIgE responses and improved clinical outcomes—contrasting with molecular spreading in the paediatric cohort. This decline was independent of allergen immunotherapy (AIT) and may reflect immunosenescence or age-related immune modulation.
Conclusions This thesis establishes new evidence linking mite molecular sensitisation patterns with clinical expression, severity, and disease evolution in AA and AR. It underscores the clinical utility of CRD for both diagnosis and prognosis and supports the need for artificial-intelligence guided stratification and development of component-based AIT in future personalised allergy care.