38541 research outputs found
Sort by
Selective laser melting of 18Ni-300 maraging steel: influence of in-process parameters and post-processing on microstructure evolution, mechanical properties, and plastic strain localisation
No full textThe use of selective laser melting (SLM) to fabricate 18Ni-300 maraging steel components for aerospace and mould making applications has attracted significant interest from both industry and academia. In particular, mould makers have utilised SLM to fabricate steel moulds with integrated conformal cooling channels, which provided enhanced cooling efficiency and reduced injection moulding cycle time compared to their conventional counterparts. However, its widespread application was hindered due to issues such as microstructural heterogeneity, mechanical anisotropy, and manufacturing defects. In the present thesis, the process-microstructure-properties relationships in SLM 18Ni-300 maraging steel were investigated to address these issues and facilitate the fabrication of additively manufactured steel moulds with mechanical properties comparable to their conventionally made counterparts.
An extensive literature review was conducted to understand how SLM process parameters (or in-process parameters) influence microstructure evolution and mechanical properties in SLM 18Ni-300 maraging steel and four other steel mould materials (i.e. H13, P20, AISI 420 stainless steel, and S136). This facilitated a novel comparative analysis of laser-powder interactions, rapid solidification, and intrinsic heat treatment across the five steel mould materials, and addressed the lack of comparative analysis specific to this topic. Complex laser-powder interactions such as the scanning motion of the laser during SLM process resulted in mechanical anisotropy and manufacturing defects, which may be mitigated via statistical process optimisation. Rapid solidification resulted in microstructural heterogeneity (i.e. variations in grain size and crystallographic texture), which was subsequently altered during post-processing heat treatment. Intrinsic heat treatment resulted in in situ precipitation hardening in SLM 18Ni-300 maraging steel and martensite tempering in the other four steel mould materials.
Experimental investigations were conducted to elucidate the process-microstructure-properties relationships in SLM 18Ni-300 maraging steel. Experiments on additively manufactured individual scan tracks and fully built samples revealed that increasing laser power (P), reducing the scanning speed (v), and reducing the hatch spacing (h) resulted in increased laser energy input (E_linear and E_volumetric) and reduced manufacturing defects. Following that, a combined statistical optimisation methodology featuring Taguchi’s methods and grey relational analysis (GRA) was implemented to determine optimal SLM in-process parameters for the multi-response optimisation of mechanical properties. Statistical analysis indicated both P and v had an approximately equal influence on the mechanical properties, while the influence of h was less significant. The mechanical properties of samples fabricated using the optimal SLM in-process parameters have high relative density (i.e. > 99 %) and were comparable with their conventionally made counterparts. Macroscale plastic strain localisation phenomena including propagation of Lüders bands and necking leading to ductile fracture in the fabricated samples were captured using optical imaging-based digital image correlation (optical-DIC).
The influence of microstructural heterogeneity (i.e. variations in grain size and crystallographic texture) on the microscale plastic strain localisation in SLM 18Ni-300 maraging steel was investigated using electron backscatter diffraction (EBSD), in situ uniaxial tensile experiments, and scanning electron microscope-based digital image correlation (SEM-DIC). Sub-micron sized speckle patterns were created via magnetron sputtering, which facilitated high-resolution strain measurements via SEM-DIC and addressed the lack of established speckle creation methodology for SLM 18Ni-300 maraging steel. Custom MATLAB scripts were employed to digitally align and overlay the EBSD and SEM-DIC datasets. This methodology facilitated a novel grain-to-grain comparison of the microscale plastic strain localisation in relation to the microstructural heterogeneity, provided grain-level insights into the role of internal misorientations on slip activity in individual grains, as well as addressed the lack of SEM-DIC investigations specific to SLM 18Ni-300 maraging steel.
Experimental findings revealed that the microscale plastic strain localisation in SLM 18Ni-300 maraging steel was driven by the interplay between microstructural heterogeneity, Ni-based intermetallics, and the impediment of dislocation motion. The densely distributed equiaxed grains contributed to grain boundary strengthening in the as-built (AB) sample. After post-processing heat treatment, the precipitation of densely distributed Ni-based intermetallics contributed to strain hardening in the solution-aging treatment (SAT) sample. The main deformation mechanism was identified as dislocation slip. Slip preferentially occurred in grains with increased internal misorientation, and intersected regions that exhibited increased kernel average misorientation (KAM). Complex slip behaviour including discrete slip, diffuse slip, and possible cross-slip (or overlapping slip) was identified in the investigated grains of AB and SAT samples. The combined kinematics of the active slip systems were reflected in the in-plane deformation behaviour of the investigated grains. The findings of the present thesis provide fundamental insights into tailoring the microstructure of SLM 18Ni-300 maraging steel for optimised industrial performance.
Finally, potential future research directions for SLM 18Ni-300 maraging steel were suggested, including numerical modelling of microstructure evolution and investigation of microscale plastic strain localisation under complex loading conditions such as multiaxial loading and fatigue
Influence factors on dynamic properties of ground improved cohesive soils
No full textDeep Soil Mix (DSM) columns are a widely adopted ground improvement method for soft, cohesive soils (alluvial, estuarine, and marine clay).
Accurate design for seismic and vibrational loading requires reliable estimation of dynamic properties: shear modulus (G), damping ratio (D), and strain thresholds. However, current studies of G and D for the treated and untreated soil were conducted separately, neglecting the critical composite effect of treated soil columns within an untreated soil matrix. Furthermore, existing dynamic regression models, primarily based on Resonant Column Test results (RCT), are limited to γ=0.1%, omitting the highly non-linear behaviour at larger strains (γ > 1%) commonly encountered during seismic events. This research addresses this gap by using cyclic triaxial tests (CT) to study dynamic properties beyond 0.1% shear strain.
The study investigates the complex influence of soil moisture content (w) / void ratio (e) in conjunction with plasticity index (PI) and effective pressure (σ’) on soil dynamic property behaviour. First, an improved basic equation model is developed for untreated soil based on PI, σ’, and w factors. This model is then seamlessly extended to cement-treated soil, incorporating cement content (CC) and moisture content (MC) to provide a unified framework that accounts for the parent soil type. In contrast, current cement-treated soil models are developed separately from untreated soil models and require complex dimensionless stress ratios. Crucially, a final composite model is developed to predict the combined effect of cement-treated soil columns in an untreated soil matrix from tests on representative cement-soil cores in untreated soil. The model, expressed in terms of the cross-sectional area replaced (AR) of untreated soil by treated cement soil, effectively simulates the non-homogeneous, anisotropic nature of the composite soil mass characteristic of DSM column-type ground improvement.
The unified models provide essential representative values for shear modulus, strain thresholds, and damping ratio to estimate the response of DSM columns under dynamic loads in seismic or cyclic loading conditions
Development of systematic algae strain and biomass detection method and database with automation engineering
No full textMicroalgae are an emerging renewable energy source due to their rapid growth and high lipid content, making them suitable for biofuel production. However, the economic challenges associated with microalgae-based biofuels necessitate innovative technologies to enhance identification, prediction, and classification efficiency, paving the way for cost-effective biorefineries. This study aimed to establish a precise, real-time, and cost-effective system for microalgae biomolecule quantification and identification. By leveraging machine learning (ML) and deep learning (DL) techniques, we successfully digitalised the prediction of C-phycocyanin (CPC) content from Spirulina platensis images. Our findings revealed that support vector machines (SVM) and artificial neural networks (ANN) achieved high accuracy when incorporating additional parameters like 'Abs' and 'Day', outperforming convolutional neural networks (CNNs). To simulate real-world scenarios, we analysed the impact of various input parameters under different lighting conditions and devices. The XGBoost meta-regressor demonstrated superior performance, offering enhanced stability and generalisation, particularly in challenging light-disturbed environments. Furthermore, our investigation into microalgae classification across three species (Chlorella vulgaris FSP-E, Chlamydomonas reinhardtii, and Spirulina platensis) showcased remarkable results. By optimising image pre-processing techniques, k-nearest neighbours (k-NN) and SVM achieved accuracies of 96.93% and 97.63%, respectively. The Azure Custom Vision model further excelled, reaching an impressive 97.86% accuracy. To advance real-time monitoring, we deployed the YOLOv8 model for microalgae detection and instance segmentation. The YOLOv8-n box detection model achieved high precision and recall, while its box instance segmentation outperformed alternatives with exceptional accuracy and reliability. Finally, to address the scarcity of high-quality microalgae datasets, we implemented generative AI techniques. Models like FastGAN, VQVAE, and DDIM effectively synthesised realistic images of Chlorella vulgaris FSP-E, Chlamydomonas reinhardtii, and Spirulina platensis significantly improving classification capabilities and dataset diversity
Bridging words and emotions: a study of emotional intelligence and emotion vocabulary knowledge in ESL learning contexts
No full textEmotions, emotional intelligence (EI), and emotion vocabulary are complex, multifaceted constructs that play a crucial role in shaping teaching, learning, and classroom interactions. Within the framework of Positive Psychology, these concepts have gained traction in education, psychology, and linguistics (see Dewaele et al., 2019; Li & Xu, 2019; Shao et al., 2020; Wang et al., 2021). While EI is extensively studied in relation to teacher wellbeing, student engagement, and pedagogical effectiveness (Corcoran & O’Flaherty, 2022; Jennings & Greenberg, 2009; Schonert-Reichl et al., 2017), an often-overlooked yet vital aspect of emotional intelligence is emotion vocabulary – the ability to recognise, articulate, and regulate emotions through language (Barrett, 2017a). Emotion vocabulary is more than a linguistic resource; it underpins emotional awareness, regulation, and expression, making it indispensable for fostering positive teacher-student relationships and emotionally supportive learning environments (Aldrup et al., 2024; Bazhydai et al., 2019; Gallingane & Han, 2015)
Despite its significance, research examining the relationship between EI and emotion vocabulary in education remains scarce. Existing studies predominantly focus on EI's relationship with burnout, teaching effectiveness, and student outcomes (Isabel et al., 2021; Mérida-López & Extremera, 2017; Molero et al., 2019; Valente et al., 2020), with little emphasis on how teachers' linguistic capacity for expressing emotions influence their EI and pedagogical practices. Furthermore, conventional vocabulary assessments largely neglect emotion-specific lexicons (Nation & Beglar, 2007; Webb et al., 2017), limiting our understanding of how teachers acquire and use emotion vocabulary in classroom contexts. Moreover, teachers’ ability to recognise, articulate, and regulate emotions significantly impacts their professional practice, classroom management, and students’ outcomes (Aldrup et al., 2024; Chen & Guo, 2020; Dewaele, Gkonou, et al., 2018; Frenzel et al., 2021; Pugh, 2008; Richards, 2020; Song, 2018; Vince, 2016). Therefore, this thesis aims to bridge these research gaps by investigating the relationship between teachers’ EI and their emotion vocabulary knowledge, with a focus on English as a Second Language (ESL) contexts.
Chapter 2 systematically reviewed existing literature on teachers’ EI and its impact on student outcomes. While meta-analyses highlight the role of EI in mitigating burnout and enhancing teacher efficacy and social-emotional competence (Isabel et al., 2021; Lozano-Peña et al., 2021; Mérida-López & Extremera, 2017; Molero et al., 2019; Valente et al., 2020), few studies explore its relationship with students’ outcomes. Findings from this chapter revealed significant methodological disparities, particularly in measuring EI and operationalising student outcomes, underscoring the need for research that specifically examines EI's role in fostering emotion vocabulary and language learning. Findings of this review also highlighted the importance of teachers' EI in fostering a positive learning environment, improving student engagement, and ultimately enhancing academic outcomes.
Building on these findings, Chapter 3 explored the linguistic gap through the lens of the affective domain, focusing on the relationship between EI and emotion vocabulary. In order to properly recognise, understand and express emotions (one of the core tenets of EI), awareness and knowledge of emotion words needs to be acquired first to accurately convey specific and nuanced emotions (Hoemann et al., 2019). Using the TEIQue-SF (Petrides, 2009) and label generation task (Bazhydai et al., 2019; Mavrou, 2021), this pilot study in a Malaysian context with 46 foundation students demonstrated a positive correlation between all facets of trait EI and emotion vocabulary. However, the label generation task produced low response variability and inadequate differentiation across language backgrounds, indicating the need for more refined, context-sensitive tools.
Chapter 4 addressed these methodological gaps by developing and validating a novel assessment tool, the Productive Emotion Vocabulary Size Test (PEVST). This tool evaluates emotion vocabulary through situated, contextualised vignettes, addressing inconsistencies in scoring methods and the lack of contextual nuance in existing measures (Barrett, 2017b; Bazhydai et al., 2019; Day et al., 2014; Goetze, 2023b; Grosse et al., 2021; Streubel et al., 2020). Unlike existing measures, the PEVST also incorporates lexical characteristics such as word frequency, a critical factor in language learning (Chen & Truscott, 2010; Schmitt & Schmitt, 2014, 2020; Wilkens et al., 2014). This study, involving 156 adult participants, revealed that word frequency, language proficiency, and emotionality significantly impact emotion vocabulary production. This study offered a novel approach to refining emotion vocabulary assessment, which does not only improve on the validity of measurement tools within emotion vocabulary research, but also revealed valuable insights into the cognitive mechanisms underlying emotion processing. Moreover, the implications of this study extend to informing targeted pedagogical strategies and curriculum development, as well as advancing theoretical understanding of the complex interplay between language and emotion.
Chapter 5 examined the relationship between teachers’ EI and emotion vocabulary using both the PEVST and label generation task with a sample of 101 teachers. As language of emotions shapes how individuals perceive and process their emotional experiences (Barrett, 2017a, 2017b), without a solid understanding of emotional vocabulary, teachers may find it difficult to even begin teaching these concepts or comprehending the emotional dynamics of the classroom; let alone recognise, understand, manage, and express emotions effectively. The PEVST shows a strong relationship between emotion vocabulary and factors such as word frequency, language proficiency, and emotionality, while the label generation task fails to establish such links. These findings not only validate the effectiveness of the PEVST but also highlight its potential as a valuable tool for ESL/EFL contexts, offering educators deeper insights into learners' emotion vocabulary use and informing more targeted, context-sensitive language instruction.
This thesis contributed methodologically and empirically to the fields of applied linguistics, education, and psychology by emphasising the role of emotion vocabulary as a linguistic and cognitive component of EI. Methodologically, in this thesis, a novel emotion vocabulary measure was developed to capture the breadth and depth of emotion vocabulary through the use of vignettes. Empirical findings in this thesis indicate that word frequency, language proficiency and emotionality contribute towards emotion vocabulary production, underscoring the closely intertwined relationship between language and emotion. The findings have the potential to inform teacher training programs, advocating for greater emphasis on emotion vocabulary training to equip teachers with the requisite knowledge and skills to meet the growing demands of curriculums and global frameworks. By highlighting the intricate connection between language and emotion, this thesis underscores the importance of emotionally aware pedagogy, offering both theoretical advancements and practical applications for improving teaching and learning outcomes
Framework for improving usability, reliability, and learning outcomes in adaptive educational hypermedia systems in higher education
No full textTraditional online learning environments generally use a one-size-fits-all approach, neglecting individual learner needs, goals, and cognitive abilities. Adaptive Educational Hypermedia (AEH) systems emerged to address this by personalizing learning content based on user profiles. The Adaptive Learning System is a computerized system developed to enhance digital interactive educational content teaching and learning. The system’s Adaptive algorithm automatically detects learners’ strengths and weaknesses through the adaptation process and provides personalized learning materials, feedback, and guidance during their learning. Adaptive content allows learners to learn the material at their own pace while providing additional support, evaluation, and resources tailored to each learner. In traditional educational systems, every student is given access to the same learning content regardless of their individual learning needs.
Higher education institutions face various challenges when testing and implementing adaptive learning systems for their domain of knowledge. Issues related to technical challenges include processing real-time data, difficulties consolidating learning concepts and solutions into existing learning management systems, and the complexity, ease of use, and usability of adaptive learning systems. However, despite the advancement and technical sophistication in adaptive algorithms, most AEH systems still face significant usability challenges. Learners often struggle with complex interfaces, limited system responsiveness, and poor adaptation to diverse demographics and subject backgrounds. Consequently, the current adaptive systems are underutilized in higher education, with limited empirical evidence on their effectiveness in enhancing student engagement, engagement, quality of experience (QoE), and academic performance.
While foundational research work (Brusilovsky, 2001; De Bra et al., 2013) and recent studies (Liu & Lim, 2020; colMOOC project) have shown the importance of adaptive educational hypermedia, yet ongoing issues persist regarding usability, learner control (Paramythis & Loidl-Reisinger, 2004), and empirical validation across varied higher education settings (Latham et al., 2004; Educause, 2018). Systematic investigations into user-controllable personalization and its effects on inclusivity and academic achievement remain limited (Xie et al., 2019; Lundqvist; Warburton, 2019). Furthermore, Li et al. (2021) highlighted that existing systems lack comprehensive cognitive modelling and fail to adequately adapt to individual learner dispositions, which restricts their intelligence and adaptability. This complexity in adaptive systems continues to pose challenges for education overall, as emphasized by Fengying Li, Yifeng He, and Qingshui Xue (2021). Addressing these issues, Dubiel et al. (2022) propose a framework for adaptive user interfaces that integrates the interaction environment to enhance system responsiveness.
Therefore, there remains a significant gap in the literature regarding the lack of empirical evaluation of user-controlled personalization, particularly in real-world, multi-disciplinary higher education contexts. Existing research has focused mainly on adaptive algorithmic and technical development, but has not given sufficient attention to practical usability (the lack of effective, usable AEH systems, learner diversity, and actual learner-centered outcomes). Moreover, there is a lack of comprehensive frameworks for evaluating the usability of adaptive educational hypermedia (AEH), and there is limited research on how these adaptive systems can foster inclusivity and fairness among diverse learner populations. Therefore, there is an urgent need to address the usability and complexity problems in AEH systems and to suggest the creation of usable, effective, personalised, learner-centric, and user-controllable AEH systems for higher education environments.
This research study addresses these gaps by introducing a personalized, user-controllable Adaptive Educational Hypermedia (AEH) framework specially designed for a Higher Education setting. It evaluates multiple versions of adaptive learning systems across multiple disciplines (English, Malay, Computer Science) revealing significant enhancements in both student achievement and satisfaction. The study not only evaluates multiple versions of adaptive systems but also offers practical recommendations for software developers and educators seeking to implement effective, more intuitive, and learner-centric adaptive systems within real higher educational environments. By exploring the relationships between learner demographics, usability, and academic outcomes, the study promotes inclusivity and equity in digital education. More specifically, it introduces a novel user-controllable personalization feature that empowers learners to customize the learning content to improve their adaptive learning experience, addressing a major limitation found in existing systems. The findings demonstrate significant improvements in student performance and satisfaction with the system’s interface, thereby enhancing subject retention, confirming that adaptive learning systems can effectively improve the academic experience in higher education environments
Hydrocarbon flame synthesis of in-situ carbon nanotubes-copper metal-matrix composite
No full textThe advance packaging of microprocessor integrated circuits (ICs) faced reliability challenges arising from failure caused by joule heating and electromigration. Impressive physical properties of carbon nanotubes (CNTs) in mechanical strength, electrical and thermal conductivity proposed a feasibilities study on producing copper-CNTs composite as solution to address the joint failure due to electromigration at the interconnect’s region. Growing CNTs directly from copper substrate remains a challenge to overcome due to low carbon diffusivity and passivation layer. This study investigates CNT growth on copper substrates via inverse diffusion flame (IDF) synthesis using acid pre-treatment to remove surface oxides and promote nucleation. Three different acids, hydrochloric (HCl), sulfuric (H₂SO₄), and nitric (HNO₃) were prepared at 1 mol/L concentration for 20 minutes. Nickel substrates pre-treated with HNO₃ served as control samples. The effects of methane flow rate (0.616 – 3.196 SLPM) on CNT yield were examined experimentally and numerically. Numerical simulations were used to predict the temperature of gas species, concentration of carbon precursors and rate of soots nucleation to investigate the correlation of methane flow rate and growth yield of CNTs. The growth control of CNTs was studied under direct current (DC) voltage bias of 0.3V – 30 V at 1.230 SLPM. Data obtained from experimental works from directional growth control serve as reference for the proposal of CNTs numerical growth models. Synthesised CNTs were characterised through various characterisation instrumentations such as Field Emission Scanning Electron Microscope (FESEM), Scanning Transmission Electron Microscope (STEM), High-Resolution Transmission Electron Microscope (HR-TEM), and Energy Dispersive X-ray (EDX). The numerical simulations of CNTs growth models would be separated into two categories, combustion of flame synthesis by ANSYS FLUENT and CNTs numerical growth models by MATLAB R2023a Simulink. Growth of CNTs was observed on nickel substrate pre-treated with HNO3 and copper substrates pre-treated with three different acids of HCl, H2SO4 and HNO3 with IDF synthesis. Both nickel and copper substrates without acid pre-treatment showed no evident of CNTs growth. Copper substrate pre-treated with HCl, H2SO4 and nickel substrate pre-treated with HNO3 showed same trend of CNTs growth yield by varying the methane flow rate with the optimum growth yield of CNTS observed to be at methane flow rate of 1.230 SLPM. Combustion numerical models were developed and validated with data from literature. The pre-treatment of substrate with acid suggested no effects on controlling growth direction of CNTs for both copper and nickel substrates for IDF configuration. The control towards growth direction of CNTs on copper and nickel substrate was achieved at voltage bias of 30V DC. The results from CNTs growth models showed close prediction to experimental data obtained in this research. These results suggested that growing CNTs on copper are promising and warrant further optimisation in growth area and more uniform distribution of CNTs diameters to satisfy minimum specification requirement for applications of ICs. This research demonstrates that with acid pre-treatment, specific IDF conditions, and voltage bias, well aligned CNTs could grow directly on copper substrates, offering a scalable pathway for future IC interconnect applications
Design and analysis of a quasigroup-based DNA encryption scheme
No full textDNA cryptography is an interdisciplinary field of cryptography inspired from DNA computing which uses DNA molecules’ role as information carrier for cryptographic purposes. In this thesis, we present an improvement on the existing algorithm with the implementation of quasigroup in the process of encryption and decryption of DNA cryptography. As opposed to traditional cryptography, which is based on numerical values, the proposed scheme makes use of DNA bases as elements of a quasigroup and unlike conventional approaches that rely solely on standard DNA bases (A, T, C, G), the proposed method introduces a DNA base U as an additional element, which appears only in the process of encryption. The encryption process involves 2 phases, namely Phase I, in which the DNA form of the plaintext undergoes transformation through a randomly generated leader and a quasigroup of order 5, and Phase II, in which the process repeats itself but the quasigroup is replaced by one of its random parastrophes. The utilisation of quasigroup operations for the proposed cryptographic scheme provides a mathematical foundation for data transformation. Notably, since the total number of quasigroups of order n increases exponentially with n, this makes them advantageous for constructing cryptosystems with extensive key space, thus ensuring enhanced security without increasing computational complexity. In summary, this thesis proposes a novel, two-phase cryptographic scheme that successfully integrates quasigroup operations with DNA encoding. The introduction of the Uracil base and the use of parastrophes were shown to produce ciphertext with near-ideal entropy, providing enhanced security against statistical attacks while maintaining linear-time efficiency suitable for larger plaintexts
Shear mechanism of reinforced concrete slabs cast with Oil Palm Kernel Shell
No full textOil Palm Kernel Shell (OPKS or commonly OPS) is one of many agricultural wastes from the extraction of palm oil. Existing disposal methods includes landfill for decomposition and incineration for energy recovery, which does not fully utilise the inherent material properties and strength of OPS. Study on granite replacement with OPS as coarse aggregate in concrete is on the rise, owing to its low density. Oil Palm Kernel Shell Concrete (OPKSC or commonly OPSC) is classified as Light Weight Aggregate Concrete (LWAC), and functions similarly albeit at a slight compromise in performance as regular Normal Weight Concrete (NWC).
The structural performance of LWAC is heavily influenced by the properties of aggregates it is batched from. Consequently, a deeper understanding on OPSC structural performance and existing design provision is necessary to aide its application within the industry. As accidents involving punching shear failure have been relatively unpredictable and sudden relative to failures in compression or flexural, studies on punching shear mechanism of flat slabs have been done under rigorous scrutiny. Therefore, the punching shear performance of OPSC as flat slabs is regarded as vital for adoption of OPSC in industrial practices.
This study investigates the punching shear mechanism of reinforced concrete flat slab cast with OPSC through experimental and analytical methods. A total of 27 flat slabs, 5 NWC and 22 OPSC, were constructed and tested under concentrated load until punching shear failure.
Mix designs for required concrete strengths were formulated which is able to achieve 100 mm concrete cube compressive strengths ranging from 15.0 Nmm-2 to 27.5 Nmm-2 for OPSC and 20.0 Nmm-2 to 25.0 Nmm-2 for NWC.
In general, collected data indicates that OPSC flat slabs performed at approximately 75.1% of their NWC counterparts under similar conditions, with ratios ranging from 66.2% to 99.4%. Punching shear resistance of OPSC flat slabs develops better than NWC flat slabs at higher concrete compressive strength, fewer rebars, steeper shear angle, and larger columns.
OPSC punching shear resistance correlates directly to concrete strength (f_c) raised to the power of 1/3. Pairing higher concrete strength with higher reinforcement ratio (ρ_1) leads to enhanced punching shear resistance. The effect of column shape (S_c) is minimal whilst punching shear resistance increases with increasing column perimeter (Γ_c). OPSC flat slabs perform less effectively at higher shear span (d_v) and but better with higher slab thickness (h).
Prediction of OPSC flat slab punching shear resistance using Eurocode 2 (EC2) NWC provisions and LWAC provisions yield a prediction ratio of 1.032 ± 0.125 and 0.800 ± 0.097 respectively. Prediction using concrete plasticity yields prediction ratio of 1.324 ± 0.305. Modified models successfully produced unity mean with standard deviations within ± 0.118 to ± 0.208 across all models.
Overall, OPSC flat slabs behaves similarly to NWC flat slabs with a slight reduction in performance in punching shear. It is confirmed that present design codes were configured to generally underestimate punching shear resistance, at about 80.0% of maximum shear capacity, even in absence of safety factors when used according to its provisions. Plastic theory further demonstrates that OPSC flat slabs perform about 83.6% of NWC flat slabs by its own provisions. Modifications proposed are recommended as a frame of reference if one were to attempt to design and adopt OPSC flat slabs in real world construction.
This study aligns with and contributes to Sustainable Development Goals (SDG) set out by the United Nations, particularly SDG 11: ‘Sustainable Cities and Communities’, SDG 12: ‘Responsible Consumption and Production’ and SDG 13: ‘Climate Action’. By promoting the use of agricultural waste as a viable structural material, this research supports the transition towards a more sustainable construction practice by providing technical support for the use of OPSC as flat slabs under punching shear load
Advancing electrochemical biosensors with AI-enhanced post-processing methods and portable electronic transducers
No full textElectrochemical biosensors (EBs) are among the most widely researched biosensing technologies due to their simplicity, cost-effectiveness, and compatibility with a wide range of nanomaterials. Their potential as point-of-care (PoC) diagnostic tools has attracted increasing attention, enabling rapid and on-site analysis for healthcare and environmental monitoring. However, the progression of EBs from laboratory prototypes to practical applications remains limited. This challenge arises mainly from inefficient post hoc analysis methods that rely heavily on manual chemometric approaches, and the lack of affordable and versatile portable transducers that are capable of supporting diverse electroanalytical techniques. Moreover, while artificial intelligence (AI) has emerged as a promising tool for data-driven analysis, its adoption in EB research is hindered by the wide range of available algorithms and preprocessing techniques, making it difficult for non-specialists to select suitable frameworks. There is also considerable uncertainty regarding whether conventional data treatment methods should be retained when deploying AI models. Additionally, the scarcity of data, arising from the complex and costly nature of experimentation, contributes to issues such as overfitting in AI models, further complicating the model development process.
To address these challenges, this thesis investigates the applications of AI, particularly machine learning (ML), to enhance EB post hoc analysis and support their transition toward practical PoC deployment. Various feature extraction methods and ML algorithms were systematically evaluated to determine optimal configurations for classification and regression tasks. The study also examined whether conventional data preprocessing methods, which are commonly employed in chemometrics, remain necessary for ML-based approaches. To overcome data scarcity and overfitting, several regularisation methods were investigated, including a bioinspired technique based on adult neurogenesis (AN), and compared against conventional approaches such as dropout, weight decay, and others. In addition, a conditional variational autoencoder (CVAE) was employed to generate synthetic EB data for data augmentation, improving model robustness and recover model performance during extremely data-scarce conditions.
The developed ML framework was validated using three case studies: (1) classification of dengue virus (DENV) serotypes, (2) quantification of acetaminophen (ACE), and (3) detection of carcinoembryonic antigen (CEA). The DENV dataset was selected for serotype classification, as secondary DENV infections can be fatal. The ACE and CEA datasets were chosen as representatives of small molecules and biomarkers, respectively, for the regression task, specifically for predicting concentrations. Concentration is critical in both cases, where excessive ACE levels indicate potential overdose, while elevated CEA levels may signal the presence of disease. For classification, the developed model achieved 100% accuracy using principal component analysis (PCA) and a multilayer perceptron (MLP). For regression tasks, the combination of discrete wavelet transform (DWT) and MLP yielded the best results for ACE quantification, achieving an R² of 0.995, and the combination of PCA and MLP again achieved the best results with R² of 0.960 for CEA detection. The findings also reveal that conventional EB preprocessing is unnecessary for ML-based post hoc analysis when appropriate feature extraction methods are applied. These optimal models were further deployed in the investigation of model regularisation. When the conventional regularisation methods show insignificant improvements, the AN regularisation method further improved the models’ performance, increasing R² values to 0.999 and 0.997 for DPV and EIS datasets, respectively. Additionally, synthetic data generation enhanced model performance by reducing mean squared error by up to 63.77% and enabled comparable performance with half the original training data size.
In summary, this thesis establishes a comprehensive framework for integrating AI into electrochemical biosensing, replacing the conventional chemometric analysis with modern data-driven intelligence. It provides systematic insights into the selection of suitable preprocessing, feature extraction, and ML algorithms for both classification and regression tasks, clarifying when conventional steps can be omitted without compromising model performance. The effective yet popular AN regularisation approach and the use of synthetic data generation via CVAE contribute new methodological directions for handling small EB datasets, enhancing model robustness and generalisability. Furthermore, the development of a smartphone-based, multi-technique potentiostat demonstrates the feasibility of translating laboratory-grade electrochemical analysis into a portable, cost-effective, and adaptable platform for PoC applications. Collectively, these contributions accelerate the realisation of intelligent, accessible, and scalable EB systems, paving the way toward autonomous biosensing and more effective data-driven diagnostics
Nutrient profiling and product innovation of underutilised crops for dietary diversity and enhancing community food entrepreneurship
No full textBackground: Malaysia’s dependence on monoculture and food imports has reduced dietary diversity and increased the prevalence of malnutrition and non-communicable diseases. Underutilised crops offer a promising alternative by improving nutrition and fostering a resilient local food system. However, gaps in knowledge, awareness, and market insight hinder their widespread use. Objectives and Methodologies: To address these gaps, this study: (1) investigates the nutritional composition and dietary potential of locally available underutilised crops through laboratory proximate analysis; (2) assesses the potential of these crops to improve the nutrient profile and consumer acceptability of food by developing innovative products and conducting sensory evaluations and surveys; and (3) evaluates the perceptions of small-scale women food entrepreneurs regarding the potential and challenges of using these crops in business settings via a pilot community intervention programme, employing survey questionnaires. Results: Nutritional profiling of 20 underutilised crops revealed significant variations in macronutrient composition, with protein content ranging from 2.3 to 26.2 g/100 g DM, crude fibre up to 42.1 g/100 g DM and energy (298.6 - 428.6 kCal/100 g DM). Clustering analysis categorised the crops into four distinct nutritional profiles: Cluster 1 (high protein), Cluster 2 (high mineral), Cluster 3 (balanced nutrition), and Cluster 4 (high fat and calorie). These nutritional variations and clusters highlight the diverse dietary benefits of these crops, demonstrating their potential to meet varied nutritional needs and address specific nutrient deficiencies in the communities. For product innovations, Clitoria ternatea (butterfly pea) flower, Lablab purpureus (lablab bean) seed, Schyzophillum commune (split gill mushroom) fruiting body, and Moringa oleifera (moringa) leaf were selected. The crops were transformed into four product prototypes, namely butterfly pea flower paste, split gill mushroom patty, moringa butter cookie, and lablab bean hashbrown, which were then assessed for their sensorial acceptance and nutritional value. Sensory evaluation showed moderate to high acceptance (6 - 7 on a 9-point scale), with moringa butter cookies and lablab bean hashbrowns being particularly well-received. The nutritional value of food products improved through both the incorporation of underutilised crops and their processing methods. Boil-blanched, steamed, and roasted moringa leaf powder showed significantly higher protein (12.1–12.2 g/100 g DM), fibre (0.4–0.6 g/100 g DM), and ash than the Control cookies. For hashbrowns, roasting increased protein by 132% (17.7 g/100 g DM), while boiling raised it by 96% compared to the Control. Roasted moringa powder in cookies resulted in reduced acceptability (7.5-24.2% lower scores across all liking attributes) but increased ash content (by 14.3%) compared to the boil-blanched powder. For the patty, the sautéed split gill mushrooms had higher lipids (by 7-9%) and energy (by 2-3%), while seared versions provided more minerals (by 5-7%). Blanching butterfly pea flowers lowered protein content (from 11.1 to 9.9 g/100 g DM) but slightly improved sensory appeal, particularly aroma (by 1.8%) and overall acceptability (by 1.9%), compared to unblanched flowers. This highlights that processing methods influenced product success, with compromises between nutrition and sensory acceptability. Consumer assessments via survey (n = 21-27; >80% aged 18-39) indicated positive consumption intent. Mean scores for likelihood to consume (3.6–3.9) were above the neutral point on a 5-point scale, approaching the "likely" anchor. Willingness to pay (mean = 3.0) approximated a "not more, not less" price point, suggesting the current formulation has market acceptance potential at a competitive price. Despite low familiarity with the specific underutilised crops (74 - 78% of respondents), affordability and accessibility were key drivers for consumer willingness to try underutilised crop-based foods. Price sensitivity strongly influenced demand (ρ = 0.754; p = 0.001), although higher-income groups showed willingness to pay premiums for value-added products. The pilot community intervention programme demonstrated significantly improved household dietary diversity, with a 38.9% increase in food group consumption, including the vitamin A-rich food groups. This improvement is consistent with the intervention's delivery mechanisms of disseminating a handbook on 60 underutilised crops, recipe sharing, and distributing ingredients. Underutilised crop consumption was still infrequent, with large variations, from 10.5 days/month for Garcinia atroviridis fruits to near-negligible levels for L. purpureus. Motivations for this consumption included liking (38%), for health benefits (13%), and for culinary use (20%). The entrepreneurs demonstrated awareness of the nutritional benefits, perceiving underutilised crops as beneficial for general health, medicinal purposes, and functional nutrition, with 75% believing that underutilised crops could improve household nutrition. However, the entrepreneurs showed limited adoption of the four developed product prototypes, with only two out of 23 businesses successfully integrating them. Despite that, most of them recognised the market potential of these products and demonstrated growing interest through the development of 22 product variations. Barriers, including consumer scepticism (40%), knowledge and familiarity gaps among entrepreneurs (30%), difficulties in preparation (30%) and product sensory challenges (20%), remain a hindrance for adoption. In a pre-post evaluation (n = 10; mean age 44.3), the food development workshop objectively improved cooking skills (by 12.3%) and awareness of underutilised crops (by 13.3%). However, it had significant changes on nutritional knowledge, and business readiness remained low due to persistent concerns over marketability and product development. Conclusion: This study demonstrates the significant potential of underutilised crops to enhance food security and nutrition in Malaysia across three critical dimensions: (1) their inherent nutritional quality, (2) innovative food product applications, and (3) entrepreneurial adoption pathways. To fully realise this potential requires the dissemination of knowledge and awareness, improving the product variations and quality, and strengthening the business pathways. These interventions will be crucial for facilitating broader adoption and maximising the impact of underutilised crops in Malaysia's food system
Access Repository Dashboard
Do you manage Open Research Online? Become a CORE Member to access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard! 👇