Search

The demand for renewable energy has increased in the recent decades due to the environmental benefits and their better availability. However, electricity generated from renewable energy sources is inherently intermittent and hence fluctuating. Suitable energy storage systems are required to stabilize the intermittency and to fill the deficit between demand and supply. Redox flow batteries (RFB) hold the capacity to meet these demands, as it offers unlimited capacity. Among different redox flow batteries, the vanadium redox flow battery (VRFB) is considered as one of the promising technologies for large-scale energy storage. The benefits of the VRFB are associated with its flexible power rating and storage time, low-maintenance cost, excellent load levelling capacity and long lifetime. The unique feature of the VRFB is the separation of the battery stack and electrolyte, which allows the decoupling of power output and energy capacity. Unlike other redox couples, due to the use of same redox species, undesirable performance fading and ion crossover can be mitigated during operation in VRFB. Simple remixing of the electrolytes in two half-cell periodically can reverse the capacity fade. Despite the ongoing commercial deployment of VRFB, there are research needs with respect to the improvement of battery performance and reducing the cost of VRFB. The research work in VRFB are mainly on membrane, electrode, electrolyte and the design of the stack, as role of these components determines the overall performance. The electrode is one of the major components of VRFB, which largely determines the efficiency and durability of the battery. In particular, as key component, electrode provides the site for redox reaction in each half-cell, which determines the cell performance. PAN-based graphite felts are widely used as electrodes in VRFB, however, pristine graphite felts from the manufacturer are inherently hydrophobic and have poor electrocatalytic properties towards vanadium species. Therefore, various modification methods are proposed to activate the surface of the graphite felt. Among these, thermal oxidation in the air is the most widely used technique owing to its simplicity and cost-effectiveness. Despite being widely used, there is a lack of scientific investigation on electrode treatment at various temperature and duration range. Most importantly, the correlation between the change in electrode physicochemical and electrochemical properties need to be investigated for achieving the optimum modification conditions. The alternative electrode modification method, which prevents the performance loss due to degradation of the electrode, also needs to be investigated. There are also no suitable methods to precisely estimate the electrolyte utilization across the porous media after the modification of the electrode by various techniques. Thus, this PhD research aimed to investigate the correlation between the physiochemical and electrochemical properties of PAN-based graphite felt electrodes. The study will use such correlation to prepare the optimized electrode with best VRFB performance and demonstrate in a single VRFB cell. With respect to the battery performance, catalyst decoration and surface modification were explored to fulfil the major objective of this study. A novel synthesis method has been carried out to dope binder-free titanium carbide (TiC) particles onto the surface of carbon fibres using titanium tetrafluoride (TiF4) as the titanium source. The method of preparation was inexpensive and yielded uniformly distributed TiC particles on the carbon fibre surface, which enhance the charge transfer kinetics of the sluggish negative redox couple (V2+/ V3+). Additionally, a systematic investigation of the effect on thermal oxidation of PAN-based graphite felts on its performance at the negative electrode of the VRFB was performed. The results from ex-situ characterization techniques were employed to identify crucial material properties, such as surface composition, specific surface area, defects and electrochemical parameters including wetting properties, mass loss, surface morphology etc. and relate with the cell performance of the cell at the specific condition. Locally resolved segmented cell study was performed to investigate the in-situ behaviour of a flow cell. A single cell with an electrode area of 100 cm2 was divided into sixteen segments to investigate the local voltage and open circuit voltage (OCV). The spatially resolved OCV was then converted to the corresponding state of charge (SOC) of the electrolyte, which was used to predict the flow behaviour and electrolyte utilization across the electrode. For the first time, a rationale to estimate the degree of electrolyte utilization is proposed based on the differences between experimental and theoretical changes in SOC. The concept was then verified with the experiment from the optimized electrode compression study of the single non-segmented cell. In addition, the study also concluded that 25% of electrode compression with respect to its initial thickness gave optimum conversion efficiency. The investigation was further extended to study the electrolyte utilization at various operating conditions, such as differently modified electrodes, various flow rates, various current densities etc. Furthermore, overpotential at each half-cell at various operating conditions was studied using the half-cell potential measurement method by inserting reference electrodes at the outlet tube of the VRFB single cell. In summary, this PhD study has successfully demonstrated the use of a low-cost novel catalyst to enhance the negative half-cell of VRFB. The study has also successfully investigated the correlation between the physiochemical and electrochemical properties of PAN-based graphite felt electrode by studying the effect of thermal oxidation on the electrode physical and electrochemical properties under various treatment temperatures and durations. By doing so, it can gain an in-depth understanding with respect to treatment temperature and duration on VRFB performance. The research has also successfully carried out first ever in-situ study based on OCV mapping to estimate the degree of electrolyte utilization. Doctor of Philosophy

Solid oxide fuel cell (SOFC) is a high-efficient energy conversion device, which can transform chemical energy into electricity directly. With the effort input by numerous researchers in the past decade, the operating temperature of SOFCs had been successfully decreased to below 500 ˚C with more than 1 W·cm-2 of output power density. At low operating temperature, one of the major issues is the sluggish kinetics of oxygen reduction reaction, which causes the major loss. To achieve high performance at a low operating temperature of 300 to 500 ˚C for SOFC, this dissertation is dedicated to developing metal-based cathodes that can meet the criteria of long-term stability, low area-specific resistance, easy to fabricate, and cost effective. To study the stability of metallic cathode, an interfacial structure characterization method, double cantilever beam delamination, was developed to determine the triple phase boundary structure for the thin-film electrode. By this delamination technique, the thermal-driven morphological evolution between the electrode top surface and the substrate contact interface were compared. For the nanoporous platinum electrode, the temperature required for significant agglomeration to occur was approximately 100 °C higher at electrolyte contact interface than at the top surface. In the next stage, inkjet printing technique was employed to fabricate Ag cathode. Comparing to the conventional thin-film electrode that was fabricated by sputtering, the cathode fabricated by inkjet printing showed improved performance and structural stability. In a 45-hour test on fuel cell operation, the cell with an inkjet-printed cathode had a current output degradation of 12.1 % at 400 ˚C; while the degradation of sputtered Ag cathode was 68.1 % after a 20-hour test. For a high-performance fuel cell, a porous silver cathode is required, and this simple inkjet printing technique provides an effective way to achieve such desirable porous structure with required thermal morphological stability. To further enhance the stability of Ag cathode, the surface modification was employed by sputtering a 2 nm-thick of gadolinium-doped ceria (GDC) on the Ag cathode. With the GDC capping layer, the thermal stability of Ag cathode was improved. The GDC-modified cathode could be operated at 400 ˚C for 24 h with 7.8 % of output degradation. To study the effect of GDC coating on surface O-exchange properties, a “porous and dense” bi-layered Ag cathode and a “porous” Ag cathode were used for surface modification. When GDC capping was applied, the ohmic resistance reduced from 1186.0 to 1052.0 Ωcm2 while the polarization resistance increased from 767.3 to 2541.4 Ωcm2 at 400 ˚C for the bi-layered Ag cathode; for the porous Ag cathode, the ohmic resistance reduced from 1190.0 to 963.5 Ωcm2 while the polarization resistance increased from 325.8 to 421.2 Ωcm2. This GDC capping layer could reduce the ohmic resistance due to the extra ion conduction pathways, while the polarization resistance increased since the GDC capping impeded oxygen surface exchange on Ag. These results provide engineering strategies for applying Ag cathode which operating at a temperature around 400 °C. Finally, the integration of Ag cathode on a micro-fabricated SOFC (or µ-SOFC) was explored. The Ag cathode penetrated the electrolyte during the operation and caused a short circuit of cathode and anode. Due to the diffusivity of Ag in the electrolyte, the design of µ-SOFC needs to be reconsidered. To employ Ag on a µ-SOFC without the short-circuit failure, further investigation on Ag diffusion-blocking layer and the thickness requirement of the electrolyte is needed. Based on these results, the opportunities and the strategies of future work are discussed. Doctor of Philosophy (IGS)

Wind power is the most successful and cost effective technology for clean energy production. Currently most of wind turbine blades are made of fiber reinforced polymer (FRP) composites. In order to provide useful information to researchers for optimized design of multidirectional FRP composite structures, it is vital to investigate the static and fatigue behaviors of FRP composites associated with failure mechanisms, which is the aim of this research. Specifically, the contents of this thesis are as follows: Firstly, the tensile and fatigue behaviors of three kinds of carbon fiber reinforced polymer (CFRP) laminates with stacking sequences of [0/+45/-45]s, [+45/-45/0]s and [+45/-45/0]2s that have been extensively used by wind blade manufacturers were investigated. Results show that the fatigue performance of the [+45/-45/0]s laminate is more excellent than [0/+45/-45]s laminate. Comparing [+45/-45/0]s laminate and [+45/-45/0]2s laminate, the predicted fatigue strength at 1 million cycles for the latter is just a little higher than that for the former. The effect of embedded delamination location on the flexural behaviors of CFRP laminates with symmetric layup sequence of [+45/-45/0]2s were investigated. Results show that the embedded delamination with the same size at different positions or interfaces has different extent of negative effects on the flexural strength of [+45/-45/0]2s CFRP laminates, but exerts no effects on the initial flexural modulus of elasticity of the laminates. After failure, all the specimens were reloaded to obtain the residual properties. It might be concluded that the reduction ratio of the initial flexural modulus is positively correlated with the areas of parallel damage (whose plane is parallel to the loading direction) and the reduction ratio of the flexural strength is almost positively correlated with the area of all the damage. After investigation of static flexural behaviors, the flexural fatigue behaviors of the [+45/-45/0]2s CFRP laminate composites were studied. Results show that the flexural fatigue life at failure can be quantitatively determined by the cycle when the tracked secant stiffness degrades by 21%. The two-parameter Weibull distribution well represents the statistical nature of the fatigue life data of the CFRP laminate. Combined with the Sigmoidal model, the Weibull analysis method reliably predicts the fatigue life as a function of the applied deflection level. Lastly, the effects of fiber orientation of adjacent plies on the mode I interlaminar fracture behaviors were studied. Results show that computed tomography (CT) scans enabled an insight into the real crack fronts of the three kinds of specimens after DCB tests. The mode I interlaminar fracture toughness at the initial stage (Ginit) is almost the same for 0//0, 0//45 and 45//-45 interfaces. While the toughness at the crack propagation stage (Gprop) in specimens with 0//0, 0//45 and 45//-45 interface continuously increases. Besides, the numerical model with variable fracture toughness is applicable to represent the mode I interlaminar fracture toughness of the FRP specimens and in good agreements with the experiments. Doctor of Philosophy (IGS)

Selective Laser Melting is a particular powder bed based Additive Manufacturing technology capable of producing metallic components from powders. There has been increasing research and industrial interests in this technology as it can achieve near-full density fabrication. Research has also shown that Selective Laser Melting is capable of producing components that are higher in strength and hardness compared to their cast counter parts due to the localised and rapid solidification that occurs during the process. As interests in Selective Laser Melting grows, more research has been focused on finding the optimal process parameters for different materials. However, current parameter optimization studies are usually carried out by trial-and-error, which is time consuming. This Ph.D. project aims to develop a semi-empirical model of the Selective Laser Melting process. This model will enable parameter optimization studies to be shortened by a providing a close estimation of the energy requirement of the process, by considering the thermal properties of the materials, the laser-material interaction and morphology of the melt tracks. The scope of this project is limited to metals and alloys as they are the most common materials used in industrial applications and some of the relevant information is available in scientific articles. A model of the process was developed and the energy requirement of the process was found to vary with the thermal conductivity values of the materials, in addition to their thermal capacities and melting temperatures. Applications and limitations of this semi-empirical model are also discussed. Through this work, optimized process parameters were also established for pure nickel and pure tin. These metals had not been successfully processed by Selective Laser Melting previously. The resultant microstructures of these materials have also been examined via various characterization techniques. Doctor of Philosophy (IGS)

The increasing demand of clean and sustainable energy drives the development of energy storage devices. Rechargeable batteries, including lithium-ion batteries (LIBs) and sodium-ion batteries (SIBs), have been considered as one of the most promising energy storage systems. However, it is still desirable to develop LIBs and SIBs with high energy storage, fast rate capability and long stability to meet their potential applications in portable electronic devices, electric vehicles (EVs) and hybrid electric vehicles (HEVs). Phosphorene, mono- or few-layer black phosphorus (BP), has attracted much attention as a promising anode for rechargeable batteries, due to its high theoretical specific capacity (2596 mAh g-1) and ultrafast Li/Na ions diffusion along the zigzag direction. However, it faces major challenges of low-yield preparation of phosphorene and the poor air-stability. In addition, phosphorene exhibits low initial reversible capacities and fast capacity fading when tested as anodes for LIBs and SIBs, due to the sluggish reaction kinetics and large volume change (-300%). Therefore, large-scale preparation methods, effective passivation approaches and advanced strategies towards improved electrochemical performance are necessary before the direct application of phosphorene into these areas. In our studies, we aim to address the problems aforementioned. Firstly, we identify formamide as an appropriate solvent to produce high-yield phosphorene with good crystallinity and tunable size distributions via liquid-phase sonication of bulk BP, which largely outperforms the commonly used ones (e.g., NMP, DMF, IPA) in previous reports. In addition, phosphorene obtained in formamide can be stable for further processing and applications. Furthermore, a densely-packed graphene-phosphorene composite (PG-SPS) is designed and prepared via a facile spark plasma sintering (SPS) method, during which densification and reduction process take place simultaneously. As a result, the PG-SPS sample exhibits much improved air-stability due to effectively suppressing the permeation of water or oxygen molecules into the sample. Importantly, the densely packed PG-SPS sample shows improved volumetric capacities due to its high packing density (0.6 g cm-3) without scarifying its electrochemical performance. In addition, to enhance the charge transfer kinetics and surface wettability with electrolyte, nanoscale surface engineering of exfoliated few-layer BP is performed by homogeneous deposition of horizontally aligned PEDOT nanofibers on surface-modified BP nanosheets. The as-obtained BP/PEDOT composites exhibit much improved wettability towards the electrolyte compared to that of the pure BP nanosheets. Accordingly, the purposely engineered E-BP/PEDOT architecture exhibits significantly enhanced reversible specific capacities compared to that of the pure few-layer BP electrode. Importantly, due to the high mass percentage of the BP in the sample, the Na storage properties of E-BP/PEDOT outperform that of the E-BP/graphene composites based on the overall mass of whole electrodes. Furthermore, a OD-2D heterostructure of Ni2P nanocrystals (NCs)-BP nanosheets (denoted as NhP@BP) was also constructed via a facile strategy. In such a heterostructure, the 2D morphology of BP nanosheets is well maintained and Ni2P NCs display an average diameter of ~ 5nm. As a demonstration of its viability, the Ni2P@BP shows much improved Li storage properties versus the bare BP when evaluated as electrode materials for LIBs. Doctor of Philosophy (IGS)

In recent years, light-emitting diodes (LEDs) enabled indoor visible light positioning (VLP) technology has attracted ever-increasing attention, due to its numerous potential applications including pedestrian localization and navigation, robots monitoring, assets tracking and even industrial manufacturing. Compared with traditional radio-frequency (RF)-based systems (e.g., Wi-Fi, Bluetooth), LED-based VLP systems has many inherent advantages, such as energy efficiency, long life-time, better positioning accuracy, etc. Besides, it can be integrated with visible light communication (VLC) system which is a fast-developing next-generation wireless communication system. Nevertheless, the development of accurate, stable and low-cost indoor VLP systems faces many challenging issues: i) limited mobility and robustness due to the line-of-sight constraint; ii) conflict of the complex processing and positioning algorithms with the limited memory, power and calculation ability of mobile devices; iii) high positioning accuracy to centimeter, even to millimeter level; iv) interruption/outage during the localization; v) the integration of VLC and VLP systems. This thesis addresses part of the pressing issues mentioned above to improve the overall performance of a VLP system. Both photodiodes (PDs) and camera/image sensors (ISs) are commonly used as the receivers of a VLP system. In an IS-based VLP system, a target’s location is calculated by solving a camera projection model which describes the one-one correlation between an object point and its projected image point. However, the conventional iterative solvers, e.g., Levenberg-Marquardt (LM), which are used to solve the projection model, are found sensitive to the initial guesses. It means the algorithm may fail to converge or suffer a long response time and a large positioning error when starting from a bad guess. To speed up the positioning process and guarantee the success rate, we derive a closed-form solution to the receiver’s position and orientation using singular value decomposition (SVD) technique and propose an SVD based positioning algorithm. Extensive tests verify that the proposed algorithm is 50-80 times faster than the conventional LM algorithm and avoids convergence failures caused by the bad initial guesses. Meanwhile, we derive the Cramer-Rao lower bound (CRLB) as the theoretical accuracy limit. Simulations are performed to reveal the impact of the various system parameters on the positioning error bound. A robust and fast 3D IS-based VLP system with centimeter-to-decimeter level accuracy is demonstrated experimentally. On the basis of the above IS-based VLP system, a sensor fusion-based positioning algorithm is proposed to enhance positioning accuracy by effectively fusing data from the image sensor and the motion sensors, both of which are widely accessible in common mobile devices. The issue of sensor fusion is formulated as a multi-objective non-convex optimization problem and mathematically solved using SVD technique, again. Experiment results show that the proposed singular value decomposition-based sensor fusion (SVD-SF) algorithm brings approximate 44% positioning accuracy improvement compared with the algorithm employing a single image sensor under the same experimental settings. Simulations are conducted to further evaluate its positioning error performance under different noises conditions. The IS-based VLP systems mentioned above requires capturing at least three LEDs in every picture to calculate the targets’ positions, otherwise positioning is interrupted or fail. The requirement greatly reduces the system robustness and flexibility. To deal with this problem, we propose a single circular LED positioning (SCLP) system where a common circular LED lamp with a point marker is used as a transmitter. The LED image is no longer treated as a point as in the existing works, but as an image whose geometric features are exploited to determine the receiver’s orientation and location relative to the reference LED lamp. The expressions for determining the receiver’s orientation and location are derived in terms of the geometric parameters. Our algorithm effectively enhances the system robustness by reducing the minimum number of required LEDs. The SCLP system is experimentally validated and centimeter-level positioning accuracy is achieved in an area of 3m × 3m. The IS-based VLP system can be classified as a self-localization system where targets detect, process and calculate the positions of themselves. This system structure may not be appropriate for portable mobile devices having limited power, memory and computation capability. Besides, the self-localization system lacks the overview of a global view of real-time locations of all targets on the map which is required in some application such as simultaneous navigation and monitoring of multiple targets. Considering these problems, a multi-target localization system is proposed for the first time in terms of visible light positioning (VLP). PDs are mounted on the ceiling to detect light signals from mobile targets (MTs) that carry LEDs, while a central server processes the received signals and calculates the positions of MTs and sends their location information back to them. By exploiting the sparse nature of a localization problem, the multi-target localization is converted into a problem of sparse matrix reconstruction. A 3-step workflow is developed to solve the problem by employing the compressive sensing (CS) theory. The proposed system and supporting algorithms are verified through extensive simulations. The impact of various parameters including the PD number, SNR, grid square size on system positioning performance is investigated. Doctor of Philosophy

Power-to-Gas (PtG) is a grid-scale energy storage technology that converts electricity into the gas fuel as an energy carrier. Specifically, it utilizes surplus renewable electricity to generate hydrogen from electrolysis with Solid Oxide Cell (SOC), and the hydrogen is then combined with CO2 through Sabatier process to form methane. The strong exothermicity of CO2 methanation remains one of the major challenges for the scale-up of the technology. This work aims to tackle the thermal management dilemma between the thermodynamic and the kinetic limitation at various conditions. Further, the economic model is built to estimate the feasibility of the large-scale deployment of the technology. Doctor of Philosophy

Due to the worldwide rapid urbanization and rising private vehicular mobility, most cities face problems such as traffic congestion, parking deficit, and cruising for parking, leading to resource wastage and environmental costs. This includes the first/last-furlong problem for personal vehicular mobility in public places i.e. the challenge to find/locate a parking space near the destination. Intelligent mobile sensing, which fuses the computing/sensing capabilities of mobile devices with machine learning, holds a great potential towards solving such urban mobility challenges. Opportunistic Mobile Crowdsensing (MCS) has also emerged as a new paradigm that enables intelligent mobility applications and services. Mobile computing research has addressed the on-street/outdoor traffic and parking problems to a great extent. However, the occupancy estimation of indoor parking garages remains mostly unsolved - especially in garages that lack vehicle-tracking sensor infrastructure and/or real-time information delivery. In fact, the major bottleneck that precludes us from reusing successful outdoor schemes into parking garages is the inadequate indoor localization potential, due to unavailability of infrastructural support such as GPS/WiFi especially when underground. In addition, enterprises using MCS face concerns regarding the scalability of their proven system deployments to newer venues; this typically calls for extra data collection for training/tuning, which can be prohibitively expensive to reach a certain spatial urban coverage required to bootstrap a viable service. Moreover, drivers who park their vehicles in such garages often face an embarrassing problem of forgetting where they parked. Most existing indoor solutions rely on infrastructure such as WiFi or BLE, whereas current smartphone-only proposals mostly require major data collection and training efforts per garage, making them impractical to adopt at indoor parking garages. In this thesis, we address these issues by proposing crowdsensing methods, implementing system prototypes and endorsing them through empirical evaluation. In the first part of this thesis, we focus on the parking occupancy inference problem for indoor parking garages without precise localization. As we cannot use mobile crowdsensing for direct occupancy counting as in the existing proposals for outdoor parking, we propose ParkGauge+ as a scalable mobile crowdsensing system to indirectly infer the occupancy of parking garages from temporal parking characteristics reported by commodity smartphones, instead of counting the parked vehicles. ParkGauge+ adopts mostly low-power sensors (accelerometer, barometer, gyroscope) to detect driving states and driving contexts and infer parking characteristics, which are utilized to infer the occupancy using Supervised Learning (SL). To achieve a scalable performance and improved spatial coverage of urban deployments, we further propose to employ Transfer Learning (TL) methods to transfer the parking occupancy inference concepts from existing ParkGauge+ deployments. This minimizes any extra data collection efforts and costs for new installations. Through extensive experiments at commercial parking garages, we evaluate and demonstrate the potential of our methods that can achieve normalized error (NRMSE) of 0.09 and 0.15 respectively for SL and TL. In the second part of this thesis, we directly address the indoor vehicular localization problem in GPS/WiFi-deprived environments such as indoor parking garages and propose a novel light-weight smartphone-only solution called ParkLoc. ParkLoc exploits the inherent planar graph structure of the navigable paths in parking facilities, to match a vehicle trajectory onto a sub-section of the map, by modeling them as sparse directed graphs. By posing the indoor parking localization task as a subgraph isomorphism problem and exploiting an approximate graph matching method, ParkLoc can track a vehicle in real-time with a median error of 4.8m and localize a parked vehicle with a median error of 2m from the nearest parking space. Furthermore, ParkLoc adopts the GraphSLAM algorithm from robotics research and applies it into the mobile sensing environment; it learns the map graph from the observed trajectory graphs and a given set of bootstrap (seed) landmark nodes, in a semi-supervised manner. A key benefit of our approach is that ParkLoc works off-the-shelf without any expensive on-site training or sensor data collection per garage. A comprehensive evaluation of ParkLoc through extensive experiments performed in 4 different garages reveals the promising performance of our graph-based approach for both localization and mapping. Doctor of Philosophy

Electrochromism is a phenomenon that materials are able to change its color reversibly under external potential. The mechanism of electrochromic materials is redox reactions in most of the cases. Composite materials have been widely used for electrochromic applications, as they have the potential to exhibit the advantages of both components. Conjugated polymer-based composite is a popular option due to the conductivity and flexibility of conjugated polymers. The challenge in design of a high-performance composite material is to achieve synergistic effects in desired aspects instead of getting average performance of the components. The design guidelines of electrochromic composite are not well established. Thus, the aim of the thesis is to design new conjugated polymer-based composite systems in which the conjugated polymers and the other components are complementary in certain aspects, and study how interfacial interactions affect synergistic effects of the complementary components in conjugated polymer-based electrochromic composites. The key hypothesis is that with proper structural design, strong interfacial interactions between conjugated polymers and the other components may induce significant synergistic effects, positively influencing electron transport, oxidation/reduction potentials, electrochemical stability, and/or mechanical robustness. To verify the hypothesis, two composite systems have been studied are molybdenum trioxide/poly(3,4-ethylenedioxythiophene):polystyrene sulfonate (MoO3/PEDOT:PSS) and polyaniline-carbon nanotube (PANI-CNT). The components have complementary properties in conductivities or mechanical properties. The interfacial interactions in each composite are electrostatic interaction and covalent bond, respectively. Both composites show enhanced electrochromic and electrochemical properties. It is proved that strong interfacial interactions between these two pairs of components induce significant synergistic effects. This conclusion provides a rationale for designing high-performance composite materials for electrochromic applications. Doctor of Philosophy

Image Classification is one of the key computer vision tasks. Among numerous machine learning methods, we choose the Extreme Learning Machine (ELM) for our image classification applications. This thesis contributes to four aspects of ELM netwroks. From the view of efficient input data, we have designed handcrafted feature extraction method for smile images classification. From the perspective of the distribution of random weights between the input layer and hidden layer, we have proposed and proved the effectiveness of the sparse binary ELM. Inspired by the deep architecture of deep learning, we have extended the single layer to multiple layers of ELM to achieve better performance on large image classification datasets. Finally, from the point of target coding, we have introduced and evaluated different target coding methods for image classification. Doctor of Philosophy

Renewable energy is fast becoming a mainstay in today’s energy scenario. Some of the main sources of renewable engery are wind, solar in addition to waves,tides,etc. These renewable energy-based production, is however inefficient from a practical as well as financial standpoint. The main reason is being the inability to forecast the exact energy that could be generated. This thesis develops a forecasting approach using interval type-2 fuzzy inferences system to address prediction intervals. The system has been adapted employing a gradient descent learning algorithm and an extended kalman filtering method. Meta-cognition is integrated into the system to improve the learning ability and prevent over-fitting. The proposed systems are used in two real-world renewable energy problems: wind and wave prediction. The wave measurement data were collected from directional waveriders deployed offshore Singapore. The experiments are then conducted on the wave energy characteristics and wind speed forecasting problems. Doctor of Philosophy

Efficient energy conversion and storage applications (e.g. liquid fuel cell and water splitting) have been identified as promising solutions to future energy depletion. However, practical installation of energy conversion and storage applications is impeded by the low performance and high cost of electrocatalysts. Therefore, it is imperative to improve the electrocatalysts’ performance so as to decrease the overall cost. The objective of this thesis is to design and synthesize electrocatalysts for energy conversion and storage applications. To this end, several works have been conducted. First, 3D Palladium nanoarchitectures were prepared, with abundant catalytic sites and open space for electrooxidation of formic acid. Second, overpotential for hydrogen evolution was lowered by fabricating two dimensional Co/NC. Third, a hybrid of MOFs was fabricated for introducing synergistic effect towards oxygen evolution. Lastly, a heterostructure of layered double hydroxides was produced for electrochemical oxygen evolution. Doctor of Philosophy

Over the last ten years, organic lead trihalide perovskites (CH3NH3PbX3, X=Cl, Br, I) have demonstrated to be the most potential material in electronic and photonic applications due to its unique properties including high absorption coefficient,1 low intrinsic recombination rates,2 ambipolar transportation,3 and long carrier diffusion lengths.4 Perovskite materials have been intensively studied for numerous applications such as solar cells,5-13 photodetectors.14-16 light-emitting diodes,17,18 and lasing,19,20 In the photovoltaic cell viewpoint, since the first perovskite solar cell (PSC) report in 2009 with a power conversion efficiency (PCE) of 3.8% only,21 the best PSC has recently achieved a surprisingly efficiency up to 23%,13 which fulfilled the commercial standard for the outside solar cell. However, the major drawback of perovskite materials is the moisture instability which limited them for being launched to the commercial market. In that critical situation, the single crystal has promised to enable the perovskite solar cells to break through their limitations. More and more reports have shown that the perovskite single crystal helped to suppress the charge recombination and prevent the bulk defect density, leading to higher PCE and longer stability.22-25 The changing of alkylammonium chains from the CH3NH3 to hydrophobic longer chains such as C6H5C2H4NH3 has been demonstrated to enhance the stability of perovskites.26 Beside the instability issue, the environmental concern about the lead content has been a consideration. However, unlike traditional PSCs based on spin coating or dip coating polycrystal thin films, the single crystal one has been difficult to synthesize due to the lack of the sufficient growth crystallization methods. This thesis shows the results for single crystals growth of large centimeter scaled 3D CH3NH3PbX3, and 2D layered (C6H5C2H4NH3)2MX3 (X = I, Br, Cl; M = Pb, Cu) perovskite single crystals by solution crystallization methods. Their crystalline, electronic, optical, phase transition and magnetic properties have been also introduced. The results should be useful for investigation of novel applications based on the stable and high-performance perovskite single crystals. Doctor of Philosophy

One dimensional nanostructure such as nanowires exhibit unique advantages such as faster charge transport, higher light absorption and high flexibility, which made them more favourable in catalysis as compared with nanoparticles. However, there were lack of mechanistic studies in one dimensional nanostructure synthesis and catalytic pathway, which results in limited morphology control and difficulties in optimization of catalysis. This thesis aims to exploit the synthesis mechanism behind catalytic-active nanowires synthesis for better morphology tunability; and catalytic mechanism of nanowires for optimization of catalytic performance. A part from that, the effect of reservoir, which was a property induced by the structure of nanowire array, was also studied for its potential usage in catalysis. Doctor of Philosophy

The current state-of-the-art lithium ion batteries (LIBs) have dominated the small format battery market for portable electronic devices， i.e., 3C products, namely, computer, communication and consumer electronics. The implementation of such technologies into automotive like hybrid (HEVs), plug-in (PHEVs), or fully battery electric vehicles (BEVs) is also quite successful. However, the realization of pure electrical propulsion requires battery materials that enable high energies to meet demands of a ∼500 km driving range, and a 10-year lifetime with driving range of 250,000 km. To achieve such a goal, the key challenge for both chemists and electrochemical engineers lies in increasing the battery energy density, while retains a long cycle life. At the moment, cathode materials remain the bottleneck due to their rather low capacity, compared with anode materials (generally graphite). Among all the common cathode materials, layered LiNixCoyMnzO2 (NCM) has overwhelming advantages for its high reversible capacity, rather low cost, good environmental benignity and high structural stability. However, the current NCM still suffers from two notable shortcomings: 1) poor rate capability especially at high C-rates due to rather sluggish Li+ diffusion kinetics; 2) fatal capacity degradation upon prolonged cycles mainly resulting from side reactions between electrode and electrolyte. Therefore, further improvements are highly desired to address these issues and satisfy the requirements for practical applications. Bearing these targets in mind, the main aims of my Ph.D. project are: 1) synthesis of NCM cathode materials with high C-rate capability and stable long-term cyclability; 2) demonstration of their practical feasibility in full cell applications, both by fabricating full lithium-ion cells with a combination of our modified cathode electrode, and via directly employing metallic lithium as anode electrode to assemble lithium metal cells (in ionic liquid electrolyte system). To address the issue of poor C-rate capability, we employed a strategy of preparation of 1D bar-like LiNi0.4Co0.2Mn0.4O2 with preferentially exposed {010} electrochemically active facets, which provide more open structure for unimpeded Li+ migration. In addition, the diffusion pathways of lithium ions are greatly reduced because of the bar shape. All these factors together contribute to the achievement of significantly enhanced lithium ion diffusivity, and thus superior high C-rate capability. Aiming at addressing the strong capacity fading issue, we adopted a surface coating strategy to stabilize the electrode/electrolyte interface and prevent detrimental side reactions. The amorphous MnPO4 coating layer acts as an ideal protective layer via physically isolating the contact between NCM active material and electrolyte. Additionally, the MnPO4 coating enhances the lithium de-/intercalation kinetics in terms of the apparent lithium ion diffusion coefficient. As a result, MP-NCM-based electrodes achieve greatly enhanced C rate capability and superior cycling stability – even under exertive conditions like extended operational potential windows, elevated temperature, and higher active material mass loadings. Through a combination of MP-NCM and commercial graphite, we further fabricated graphite/MP-NCM full lithium-ion batteries, which show remarkable long-term cycling stability (capacity retention ratio up to 74.1% over 2000 cycles at 1C), and deliver a high gravimetric energy density of 376.2 Wh kg-1, demonstrating great practical feasibility. Employing an intrinsically safer electrolyte system, ca., ionic liquid as alternative, the as-fabricated lithium metal batteries allow for outstanding cycling stabilities with a capacity retention of well above 85% after 2000 cycles. Currently, the commercial electrode fabrication generally employs a large portion of toxic components and processing solvents (i.e., NMP), particularly for cathode electrodes. To design greener LIBs, in this thesis, we also dedicated to replacing the conventional organic PVDF-based binder with aqueous CMC-based binder and investigating its effects in electrochemical performance. Doctor of Philosophy

A thermally comfortable building indoor environment is imperative for human health and work productivity. Although the major energy consumption in a tropical building is attributed to its Air-Conditioning and Mechanical Ventilation (ACMV) cooling systems, yet complaints of thermal dissatisfaction is prevalent among occupants. Prediction of the occupant’s thermal comfort can be instrumental in bridging this gap by utilizing the predicted thermal state (Cool-Discomfort/Comfort/Warm-Discomfort) as a control criterion for the ACMV system. However, the existing thermal comfort prediction methods rely primarily on environmental information. We envision that a possibly simpler approach would be a direct investigation of the human physiological parameters, since the bodily responses are direct consequences of surrounding thermal ambiance. In this study, we investigate the potential of six different physiological parameters to predict thermal state namely, skin temperature, skin conductance, pulse rate, blood oxygen saturation, and systolic/diastolic blood pressure. Human subject experiments were conducted, during which these six physiological responses and four subjective responses (thermal comfort, thermal preference, humidity sensation, airflow sensation) were recorded in conjunction with a thermal sensation survey while environmental conditions were varied. An overall relational study of all the physiological, subjective, and environmental parameters is performed to gain a clearer understanding of human thermal comfort. An index termed as Thermal State Index (TSI) is introduced to quantitatively represent thermal comfort. A study of skin temperature of the selected hand skin location revealed that it strongly correlates with the overall thermal sensation. It is verified that the rate at which skin temperature changes carries significant information about the thermal state. However, considerable individual differences are revealed. We develop a novel normalization process to address both inter and intra individual differences by incorporating body surface area and clothing insulation, respectively. Based on this normalization process, we develop a TSI predictive model named Predicted Thermal State (PTS) model. While non-normalized skin temperature alone could predict only about (60-65)% of thermal states, the PTS model based on the normalized skin features could predict accurately 87% of thermal states. The combination of skin temperature and its gradient carried more predictive potential than the skin temperature alone. Therefore, the temporal profiles of skin temperature could portray adequate information without normalization. Based on these temporal profiles, we develop another TSI predictive model named the Temporal Profile based Predicted Thermal State (TPPTS) model. A 2-D sensor data to visual space domain-transformation process is developed. Based on a deep convolutional neural network, the TPPTS model achieves 93.33% accuracy. A study of pulse rate response also revealed considerable individual differences. A novel normalization process using logic gate operations is developed to address the differences with gender and BMI as individual factors. Additionally, a new potential feature of pulse rate is extracted using a spectral analysis method. Based on these features and novel processes, we develop another TSI predictive model named the enhanced Predicted Thermal State (ePTS) model; it achieves about 97.15% accuracy. A collective study of all six physiological and four subjective responses revealed significant gender differences in both responses. Hence, a separate comfort study for each gender-group is conducted, where potential predictive features are identified. Derivative features such as change rate and mean squared gradient are also studied. Based on the identified features and a Random Forest algorithm, we develop another TSI predictive model named the Gender based Predicted Thermal State (GPTS) model; it achieves 92.86% and 94.29% accuracies for males and females, respectively. A canonical correlation analysis reveals strong relations between physiological and subjective responses. The four proposed TSI predictive models significantly outperform the conventional PMV method. A comparative analysis of the proposed and existing methods is performed. Subsequently, we propose a novel framework for indoor climate control that can provide optimum comfort with minimum energy consumption of the ACMV system. It incorporates a TSI predictive model and a novel optimization algorithm named the OAT algorithm, that provides the optimal operating condition based on the current TSI. The framework exhibits 36.5% energy saving potential. Additionally, we study four less-explored non-physiological parameters namely, mean radiant temperature (MRT), age, gender and outdoor weather in context of their significance for thermal comfort. As for MRT, we investigate the effect of the common practice of assuming it to be equal to air temperature on comfort assessment, and subsequently develop an alternative model (R2 = 0.89) to estimate MRT. Age, gender, and outdoor effective temperature, which are not included in the existing comfort models, are found to be important factors for thermal comfort. We believe that the findings of this thesis will assist the built environment community with a better understanding of human thermal comfort and associated phenomena. Doctor of Philosophy

Solving optimisation problem is computationally demanding, and hence Model Predictive Control (MPC), an optimisation-based control technology, is traditionally employed in applications with slow dynamics. In recent years, due to its ability to systematically handling constraints and multiple-input and multiple-output systems, MPC has been extended to many non-traditional areas, in particular networked or embedded applications. However, limited computational resources poses challenges for embedded implementation of MPC. Computational resources to solve MPC problem may be time-varying and insufficient at times. Moreover, the measurements transmitted through a communication network may be unavailable due to network congestion or packet dropouts. To enable MPC be more widely used in networked or embedded applications, this thesis addresses two aspects: (1) methods to efficiently solve the optimisation problem in embedded platforms and (2) methods to tackle time-varying computational resources. System of linear equations with the saddle point type is the main computational load in some MPC-tailored algorithms. Existing MPC-tailored algorithms have not exploited banded null bases in the sparse MPC formulation. In this thesis, an algorithm for exploiting banded null bases is presented. The proposed algorithm is tested with a wide range of MPC benchmark problems. Implementation results on an industrial embedded platform confirm that significant reduction in the runtime for the Alternating Direction Method of Multipliers (ADMM) can be achieved by the proposed algorithm. Methods for improving the conditioning of the banded bases are also developed. Event-triggered Sequence-based Anytime Control (E-SAC), recently proposed in the literature, can handle the time-varying computational resources effectively. The main idea of E-SAC is, when computing resources and measurements are available, to compute a sequence of tentative control inputs and store them in a buffer for potential future use. Existing E-SAC in the literature only features one control law in the buffer. In this thesis, E-SAC is extended to schemes featuring multiple control laws. Numerical simulations show that performance in terms of empirical closed-loop cost, channel utilisation and regions for stochastic stability guarantees could be improved, compared with the existing E-SAC. However, the current stability analysis framework for E-SAC, the State-dependent Random-time Drift approach, becomes combinatoric and difficult to use when extending to the multi-control law E-SAC. To overcome this, a new stability analysis method for E-SAC based on Markov jump system is developed. Using the proposed stability analysis method, stochastic stability conditions of existing E-SAC are also recovered as a special case. In addition, the proposed technique systematically extends to other more sophisticated E-SAC scheme for which, until now, no analytical expression had been obtained. Doctor of Philosophy

Solar PV is renowned all around the world, but its prevalence varies with a country’s solar profile. Singapore is an island country which has solar energy as the only promising renewable energy option. On the UN’s request to submit the Intended Nationally Determined Contribution (INDC) against carbon emissions, Singapore, as a promoter of efficient energy, has come up with an elaborate Climate Action Plan towards carbon emission reduction. An aspiring leap by Singapore towards 350MWp PV capacity by 2020 is on trial, which would involve numerous PV installation projects. The thesis at one end demonstrates the quantitative impact of PV installation on long-length transmissions, while at the other end, it showcases the use of intelligent algorithms for forecasting future PV power output. Chapter 2 deals with a 3.3MWp PV installation in the low voltage side of the Jurong Port distribution grid, providing comprehensive steady-state analyses in different loading scenarios. Comparison of different configurations of PV and ESS is done and it is found that the PV installation improves the bus voltage profile, reduces line congestion and circumvents substantial transmission losses. Accurate predictions of solar power are important to the grid operator for ensuring energy management from multiple sources without jeopardizing stability and to the PV plant owner for scheduling plant maintenance periods and avoiding penalties imposed by the grid operators due to power imbalance costs. It is evident that meteorological data like solar irradiance is more readily available than historical PV power output series with hourly samples. In this case, indirect forecasting can be utilized where solar irradiation is predicted first, which is followed by obtaining the PV power output using a PV performance model of the plant. Since PV series during the day experiences lots of variations, point forecasting can be quite uncertain. The inherent uncertainty in the point forecasts can be quantified by associating them with a probability distribution to form prediction intervals (PIs) which is a more interpretable representation of uncertainty. Chapter 3 presents a probabilistic forecasting approach using a nonparametric PI formation method based on Extreme Learning Machine. No prior assumption on the error distribution is required for the PI formation. Solar irradiance data from NUS geography weather station, Singapore, is analyzed and assembled into two separate sets for better model performance. Cross-validation and Grid search followed by Differential Evolution are utilized to tune the hyper-parameters of the proposed model. Coverage probability and interval scores are evaluated for the resulting PIs which show promising results. Master of Engineering

Non-recurring incidents such as accidents, vehicle breakdowns, etc. are leading causes of severe traffic congestion in large cities. Consequently, anticipating the impact of such events in advance can be highly useful in mitigating the resultant congestion. However, availability of partial information or ever-changing ground conditions makes the task of forecasting the impact particularly challenging. In this thesis, we propose adaptive ensemble models that can provide reasonable forecasts even when a limited amount of information is available and further improves the prediction accuracy as more information becomes available during the course of the incidents. Furthermore, we consider the scenarios where the historical incident reports may not always contain accurate or complete information about the duration and length of congestion due to the incidents. To mitigate this issue, we fi rst quantify the effective duration and queue-length of the incidents by looking for the change points in traffic state (average traffic speed and flow data) of the individual upstream links and then utilize this information to predict the duration and queue-length of the incidents. The prediction models forecasts the values continually with elapsed time until the end of the incidents to comprehend the temporal dynamicity. We compare the performance of different traditional regression methods in predicting the duration, and the experimental results show that the Treebagger outperforms other methods. The overall MAPE value averaged over all incidents improves by 50% with elapsed time. On the other hand, we build a queue-length prediction model using a Long Short-Term Memory neural network which incorporates the updated traffic data and various spatio-temporal features as inputs. At the start of the incident, the proposed model has a mean error value of 73.7%, which reduces to 45.6% after one hour of prediction. Moreover, we are interested in not only predicting incidents' impact but informing the drivers about the current and future traffic state is also our area of concern. Currently, in order to inform or alert the commuters about the traffic situation during the road incidents, several LED displays, better known as variable message signs or VMS messages, have been installed by the LTA on the expressways of Singapore. Therefore, apart from building the predicting models, this thesis also aims to evaluate the impact of VMS displays on the overall traffic distribution of Singapore whether these displays are really helpful to the drivers or not. To this end, the incidents data and their corresponding VMS messages are collected from the two busiest expressways of Singapore, namely Pan Island Expressway (PIE) and Central Expressway (CTE). The analysis shows that approximately 14% of the vehicles change their direction after the VMS messages have been activated. Lastly, since Singapore is a tropical country having a significant amount of rainfall throughout the entire year, the weather condition has a noticeable impact on the traffic of Singapore. Therefore, we also aim to analyze the influence of rainfall on traffic incidents if the frequency of these incidents, especially accidents, increases after rainfall or not. Therefore, the rainfall data acquired from the National Environmental Agency (NEA) of Singapore is analyzed to investigate the correlation between the occurrence of traffic incidents and rainfall. Overall, the obtained results support the hypothesis that the frequency of traffic incidents is higher during rainfall as compared to dry periods. Moreover, the frequency is the highest after rainfall. Besides, it is observed that traffic speed and flow decrease by 10.14% and 3.88% respectively during rainy weather. Overall, this thesis aims to help avoid incident-induced traffic jams by designing the urban traffic prediction models. Thus, the cost of time and fuel can be saved, which will benefi t the national economy as a whole. Doctor of Philosophy

Time-domain mooring simulation is a direct and effective method of evaluating mooring line loads and floater motions, in which nonlinear effects including drag forces and seabed interaction are readily accounted for. The formulation of seabed models affects the accuracy of line loads, including maximum tension as well as fatigue results, and also floater motions in coupled simulations. The interaction between the seabed and discrete line structural models is challenging in time-domain numerical simulations and a myriad of seabed models have been proposed in the literature. However, issues such as the selection of suitable seabed force coefficients and the application of suitable line discretisations, to achieve a balance between accuracy and computational cost, remain challenging problems to be resolved. This thesis investigates the shortcomings of existing numerical methods in accurately capturing the dynamic mooring line-seabed interaction and proposes new solutions. The line-seabed interaction problem is pertinent to catenary mooring lines which experience liftoff from and grounding on the seabed when undergoing large dynamic motions. This interaction is accounted for by various seabed models and it is known that the action of liftoff and grounding may lead to large dynamic tension fluctuations. These fluctuations may be spurious due to the inability of discretised mooring models to adequately account for the effect of the seabed on the mooring line. The effect of line discretisation and seabed model formulation on the tension fluctuations is investigated using the widely used spring-mattress approach. An in-house mooring code was developed to perform these investigations. For code validation and benchmarking, and to illustrate the existence of the tension fluctuations problem due to nodal grounding in existing mooring line simulation codes, comparisons are made to a commercial software. This work aims to contribute to the field of numerical mooring line modelling by identifying the root cause and highlighting the conditions leading to the production of the large dynamic tension fluctuations that tend to occur due to impact loads imparted by spring-mattress type seabed models on discrete line structural models. In particular, these tension shock waves are found to be caused by strain discontinuities manifesting in the vicinity of the touchdown zone as a result of discrete line elements and nodes coming into contact with the seabed. The likelihood of occurrence of the strain discontinuities is determined to be dependent on the impact speeds and orientations of line elements in relation to the seabed; higher impact speeds and close alignment between the seabed normal direction and the axis of line elements increase the tendency of occurrence of the strain discontinuities. In light of these findings, a novel seabed reaction force model that inhibits the production of strain discontinuities in a discrete line model is proposed. The results from this research further show that using a suitably fine discretisation for the mooring line model, while incurring a higher computational cost, is a generally applicable approach to improve result accuracy. Specifically, the modelling of line touchdown-liftoff effects requires a finer mooring line discretisation at the touchdown point to avoid the production of spurious line tension fluctuations. The numerical experiments performed in this research demonstrate that, for the purpose of ameliorating the effects of propagating stress waves arising from rapid line grounding, a hybrid discretisation employing fine elements within a limited span of line close to the touchdown zone and coarse elements elsewhere is sufficient and preserves the accuracy of predicted peak tensions. The results of the discretisation refinement study in this work further suggest that specifying a numerical chain element size close to the physical chain link size within the locally refined zone is sufficient even under severe excitation conditions which promote the production of the tension fluctuations. The disparity in element sizes within and outside the local refinement zone gives rise to a stiff dynamical system. This work introduces an approach for applying adaptive discretisation to the line structural model with a non-uniform mesh, and dual-rate time integration for the resultant stiff dynamic system. The results in this research suggest that a dual-rate time integration approach for the proposed hybrid spatial discretisation of the line structure significantly reduces computational times for smaller refined zone line spans as a proportion of total line length. As the touchdown point changes with time, the proposed adaptive discretisation procedure enables the locally refined zone to shift in tandem in order to limit the spatial extent of the refined domain, thereby preserving the computational efficiency gains of the dual-rate time integration scheme even as the touchdown point changes significantly. Hence, in this context, this thesis presents a method to achieve high accuracy at lower computational expense. Future studies can extend the proposed method to cable models with additional degrees of freedom such as bending and torsion. Doctor of Philosophy

Solid oxide electrolyser cell (SOEC) is regarded as a promising device for future energy storage and for the deduction of carbon dioxide emission. This thesis aims at finding ways to help improve the long-term stability of SOEC from theoretical perspective. With the background and motivation discussed, we focus on improving the cell performance and seeking for better understanding in the electrolyte materials, especially in the electrolyte material yttria stabilized zirconia (YSZ) and gadolinium doped ceria (GDC). We find that the maximum ionic conductivity of YSZ may come from the competitive interaction between the positive effect from the increase of oxygen vacancies and the negative effect from the increasing of yttrium ions. We also find that oxygen ion diffusion is hindered by the existence of GB in YSZ. With further investigation, we find that the oxygen ions diffuse faster along the GB interface than across it. The results suggest that GB shall be avoided in synthesizing if possible, and that the direction of GB interface shall be aligned parallel to the ion conducting direction to improve cell performance. A minimal layer thickness requirement is found on the thickness of the GDC interlayer, which blocks reactions between electrolyte and electrode. The minimal thickness is found to be ~10nm and ~2µm under 700℃ and 1000℃, which provides useful advices for researchers in the designing and synthesizing of SOEC components. A relationship between the lattice parameter and overall thickness of CeO_x laminated structure is given as a tool for researchers to estimate the oxygen conducting property based on the thickness of GDC interlayer. Electronic structures and work function in ceria (CeO_2) and GDC are studied. Doping of gadolinium significantly decreases the band gap of ceria, while thin film structure of GDC raises up the band gap a lot. The existence of oxygen vacancy is found to be responsible for the decrease in band gap and helps explain the electronic conductivity of GDC under low oxygen partial pressure. In conclusion, the results of this thesis help improve the understanding in SOEC electrolyte materials and provide useful information and advices to researchers who are interested in designing and making SOECs with improved performance. Doctor of Philosophy

This thesis deals with colloidal synthesis of hybrid organic-inorganic halide perovskite nanocrystals (NCs) for light emitting diode (LED) applications. The influence of, variations in the precursor ratios, as well as NC size distribution modulation via solvent engineering, during colloidal synthesis of methylammonium lead bromide [CH3NH3PbBr3] perovskite NCs, on the efficiency and stability of the perovskite NC-based LEDs are investigated. A comparison of the CH3NH3PbBr3 and formamidinium lead bromide [CH(NH2)2PbBr3] NC synthesis optimizations for LED applications is also presented. The insights obtained from this research are expected to be useful for the design of perovskite NC synthesis protocols for optoelectronic applications. Doctor of Philosophy

Global energy and environmental crisis have received enormous research attention for a long time. Fossil fuels as the primary energy source, comprised 86% of global energy consumption in 2015. However, the consumption of this nonrenewable resources brings about not only the environmental problem such as global warming, but also the economic problem since fossil fuels’ price increased rapidly due to the limited recoverable reserves. With the inexhaustible irradiation from the Sun, solar energy is agreed to be the most promising renewable energy resource. Inspired by Mother Nature, converting solar energy to chemical energy has been intensively studied. Systems including photocatalytic and photoelectrochemical water splitting have been developed to fulfill the ultimate goal of high efficient solar-to-hydrogen conversion. The overall objective of this interdisciplinary research program is to improving the overall solar-to-hydrogen efficiency by rational design of catalytic nanomaterials. Firstly, a new kind of photosensitizer were studied for improved light harvesting. A series of MOFs based on MIL-125 have been demonstrated as promising photosensitizers for PEC water oxidation. When applied MOFs/TiO2 heterostructure materials as the photoanodes for solar water oxidation, the photocurrent of TiO2 could be improved by nearly 100% under visible light through sensitization with aminated Ti-based MOFs. Incident photon-to-current conversion efficiency matches well with the UV–vis absorption of modified TiO2 photoanodes, which confirm the sensitization of MOF materials. Highly porous structure of MOFs enables further tuning of light harvesting efficiency. By coupling with plasmonic Au nanoparticles, the light absorption could be further improved due to the LSPR effect. Secondly, a new noble-metal free HER catalyst were synthesized for improved reaction kinetics. A general one step method to prepare ultra-small transition-metal carbide nanoclusters encapsulated in N, S co-doped graphene were developed, and used for all pH hydrogen evolution reaction. The as-prepared MoC1-x/C materials possess ultra-small particle size (2.8 nm) with narrow size distribution (1.8 to 3.2 nm), which provides lots of catalytically active sites. The intimate interaction between transition-metal carbide nanoclusters and N, S co-doped graphene modifies the binding strength of protons on transition-metal carbides. The porous graphene supports enable quick charge transport and fast reactant/product diffusion. All of which contribute simultaneously, giving rise to the excellent HER performance of the transition-metal carbide/graphene composite catalysts. Finally, photoelectrochemical reforming of biomass-derived alcohol oxidation has been developed for decreased overall energy consumption in solar-to-hydrogen system. In this new system, pure hydrogen and value-added chemicals were simultaneously produced driven by solar energy at room temperature. The hydrogen quantum efficiencies reach 80% across the entire visible region of the solar spectrum, leading to an ultrahigh hydrogen production rate of 7.91 μmol min-1 cm-2 illumination area of a single-junction GaAs solar cell under AM 1.5G illumination. Moreover, by properly choosing the anodic catalyst, alcohols can be selectively converted to useful liquid chemicals with nearly 100% selectivity. Doctor of Philosophy (IGS)

Tremendous attention has been made to the development of high energy and power lithium-ion batteries (LIBs) used in electric vehicles. Unfortunately, existing commercial graphite anode material has inadequate high energy and power density. Recently, Li3VO4 (denoted as LVO) anode has been reported as a promising insertion type material to replace graphite due to theoretically higher capacity (395 mAh g–1) and cycling stability. Furthermore, LVO has an appropriate voltage window of 0.5–0.8 V vs. Li+/Li. This allows greater security since safety issues are commonly associated with the formation of Li dendrites that happens below 0.1 V vs. Li+/Li. Apart from LVO, vanadium sulfide VS4 is another promising anode material with a high theoretical specific capacity of 1196 mAh g-1. The main objective of this thesis is to investigate the morphological effect, along with the effect of compositing LVO and VS4 with carbon on electrochemical performances. Doctor of Philosophy

The large-scale implementation of renewable energy technologies, e.g. solar photovoltaic and wind turbine, requires effective and stable electric energy storage systems, due to the inconstancy of such renewable energies. Among some potential candidates, the vanadium redox flow battery (VRFB) is promising to play such a role, thanks to its advantages: power and energy capacity can be independently designed; the simple structure of cell and stack; quick response and long cycle life. One of the challenges in VRFB development is to improve the low solubility of V(V) species in sulfuric acid. The positive electrolyte of the VRFB consists of a mixture of vanadium salts in oxidation states 4+ (IV) and 5+ (V), dissolved in sulfuric acid, with a total vanadium concentration of typically 1.5 – 2 mol.dm‒3. As the battery is charged, the ratio of V(V) to V(IV) increases. Therefore, the VRFB is generally operating with the positive electrolyte in a metastable state. Consequently, there exists the risk that V(V) ions may condense to form V2O5 and then followed by the precipitation at high temperatures, which is almost irreversible in the charged electrolyte. In practice, the kinetics of precipitation is strongly dependent on temperature and state-of-charge (SOC). By careful control of temperature and state-of-charge, VRFB manufacturers avoid precipitation. However, these measures add to the cost and can reduce performance. Therefore, much effort has been made to find electrolyte additives that would allow wider operational temperature windows. The most common stabilizers to date are H3PO4, (NH4)2SO4, and HCl. It has been suggested that halide ions can form complexes such as [V2O3Cl2.6H2O]2+ with the hydrate pentacoordinated [VO2(H2O)3]+ cation, in which the chloride hinders the polymerization to V2O5. However, halide in the electrolyte gives the risk of forming poisonous halogen vapors, while H3PO4 and (NH4)2SO4 are rather limited in their effectiveness. A large number of organic stabilizers have also been suggested, but little has been reported as to their long-term stability. Therefore, in this thesis, a systematic development strategy (precipitation test, electrochemical test, material characteristics, and etc…) was conducted for finding out an electrolyte formulation (especially new additives) for a vanadium-based electrolyte, which is halide-free and stable to high temperatures. The electrolyte is also expected to demonstrate the longer-term stability and suitability for laboratory-scale VRFB and eventually for a 1 kW system. In this Ph.D. thesis, we summarize the progress and achievement in the development of novel electrolyte formulation for the VRFB. It is demonstrated that the novel thermally stable additives have been successfully developed for the VRFB electrolyte, through the combination of one inorganic component and one organic component, which could provide the co-stabilizing effect to prevent the precipitation of positive vanadium electrolyte at high temperatures. The thermal stability of pristine vanadium electrolyte has been improved nearly 4 times by the addition of combined additives. In addition, as-developed combined inorganic-organic additives also could maintain the rate of V2O5 precipitate below the value of ~10 mol.% of V(V). By contrast, more than 60 mol.% of V(V) in the pristine electrolyte was precipitated within 7 days at 50 ºC. Since the recipe is halide-free, the risk of toxic halogen vapor formation has been completely eliminated. Besides that, no other harmful contaminants were released under the operation of new electrolyte formula. So-developed halogen-free thermally stable electrolytes also satisfied the longer-term stability and suitability in a practical VRFB cell at high temperatures. The electrolyte formulas with the presence of combined inorganic-organic additives could preserve the performance of a laboratory-scale VRFB at 50 ºC for over 100 cycling number. It also preserved the performance of a large-scale 3-stack VRFB system. Although there was a minimal drop in the efficiency of the cell, the drop could be diminished by varying the addition amount of additives or using suitable ion exchange membrane. Moreover, owing to its high thermal durability, the VRFB could be operated at relatively high temperatures without the necessity to install a cooling system, resulting in higher efficiency and lower cost. The innovative vanadium electrolyte developed in this Ph.D. study is highly promising to be used in commercial VRFB system. Doctor of Philosophy (IGS)

This study focuses on reactive magnesium oxide (MgO) cements which has lower energy requirements and CO2 emissions compared to common Portland cement. A very limited amount of work has been reported on the carbonation degree, performance and durability of MgO cement concrete under various conditions. The main goal of this research is to (i) enhance the carbonation, mechanical performance and microstructural development of MgO cement concrete containing different additives, (ii) evaluate the stability of the formed phases under various conditions involving high temperatures and aggressive environments, and (iii) provide a comparison between commercial and lab scale blocks in terms of performance and microstructure. In all studies, the performance and microstructure of reactive MgO cement were investigated by various methods. The presented findings did not only facilitate the understanding and controlling of the properties that affect the performance and microstructure of MgO cement-based mixes, but also demonstrated the potential of commercial MgO cement-based concrete. Doctor of Philosophy

Solid oxide electrolyzer cells (SOECs) which serve as a highly efficient and viable technology for energy storage, have drawn great attention in the recent years, especially with the continuous rise of renewable energy usage. However, conventional SOECs using state-of-the-art materials are sensitive to contaminants and require ultra-pure feedstock gas which increases their operating cost significantly. Therefore, it is desirable for SOECs to utilize cheaper and more abundant resources such as seawater or industrial flue gas as feedstock gas. This thesis investigates the direct use of seawater electrolysis, the development of a sulfurtolerant cathode/fuel electrode and the modeling of SOECs based on simulated flue gas environment in the cathode. State-of-the-art SOECs are studied for direct electrolysis of water vapor produced from seawater using button cells with Nickel-(Yttria-stabilized-Zirconia) (Ni-YSZ) cathode. Using an ion chromatograph mass spectrometry, water vapor produced from seawater is found to be free of the contaminants, which are present in the seawater. Current-voltage curves, AC impedance spectroscopy and galvanostatic tests are used to evaluate the electrochemical performance and durability of the cell. Similar cell performance is observed using water vapor produced from pure water and seawater. Their short-term degradation rates are similar, which are registered at 15% 1 OOOh- 1 for both cases. A button cell with its cathode impregnated with sea salt is used to investigate the effects of direct sea salt contamination in SOECs. Both the uncontaminated and contaminated cells exhibit rather similar performance as observed from the current-voltage curves and impedance spectra. The difference in area specific resistances between the uncontaminated and contaminated cell are all within a 10% range. Rather similar short-term degradation rates of 15% 1000 h-1 and 16% 1000 h-1 are recorded for the uncontaminated and contaminated cells, respectively. Post-mortem analysis using scanning electron microscopy and energy-dispersive X-ray spectroscopy show that the sea salt impregnated into the cell has been vaporized at a typical SOEC operating temperature of 800°C over the period of operation. Lao.1sSro.2sCro.sMno.sOJ-.S (LSCM)-based composites are studied as the cathode of SOECs for direct co-electrolysis of water vapor and carbon dioxide in industrial flue gas. The electrochemical behaviors of the composite cathodes are investigated using three-electrode electrolysis cells with Pt as counter and reference electrodes. Electrochemical results from current-voltage curves and AC impedance spectroscopy among pure LSCM, LSCM(Gadolinia-doped-Ceria) (GDC), LSCM-YSZ and LSCM-(GDC-YSZ) have shown that LSCM-GDC exhibits the highest water vapor electrolysis performance. The ratio between LSCM and GDC is further optimized and it is shown that the LSCM-GDC with 50-50 wt.% for each component exhibits the highest performance. A 60-40 wt.% Ni-YSZ state-of-the-art cathode is used to benchmark with the optimized LSCM composite cathode. It is shown that the optimized LSCM-GDC cathode exhibits better performance for water vapor electrolysis. Under a cathodic current of -0.1 A cm-2, the optimized LSCM-GDC cathode shows much slower degradation, about 10 times slower as compared to the Ni-YSZ cathode when exposed to 10 ppm of sulfur dioxide for up to 72 h. The optimized LSCM-GDC cathode also shows promising performance for co-electrolysis of water vapor and carbon dioxide at high current densities and stable performance under simulated flue gas condition with 5 ppm of sulfur dioxide and 6.5% oxygen in the feedstock gas. Further impregnation of and Palladium by infiltration into the cathode shows an overall increase in co-electrolysis performance. A two-dimensional SOEC model using COMSOL Multiphysics modeling software is used to investigate the production rates of hydrogen and carbon monoxide as well as the gas molar distributions in the cell during operation under flue gas condition, where the results are being compared with pure co-electrolysis of water vapor and carbon dioxide. Under a certain bias, it is found that the hydrogen and carbon monoxide molar fractions increase along the cathode channel in both cases, but the rate of increase is much slower in the case of simulated flue gas condition as the feedstock gas. This is due to the chemical reaction with oxygen which is present in flue gas, which causes their production rates to be lower than the case of pure water vapor and carbon dioxide. The molar fractions of water vapor and carbon dioxide are of the opposite, where they decrease along the cathode channel, but less significantly in the case of simulated flue gas condition. The oxygen molar fraction decreases along the cathode channel and increases more significantly along the anode channel under simulated flue gas condition. Doctor of Philosophy (IGS)

Traffic and transportation are crucial to the sustainable development of most metropolitan cities, where the stochastic shortest path (SSP) problem for vehicle routing is a challenging topic. For this problem, optimization-based methods and multiagent-based methods are usually adopted to compute the optimal path for a single vehicle and multiple cooperative vehicles, respectively. Most of these methods take into account various uncertainties in traffic and yields route recommendations dynamically, leading to a better solution to the development of sustainable transportation systems. At the same time, various criteria for the optimal paths have been proposed for the SSP problem. Among them, the probability tail (PT) model aims to find a path that maximizes the probability of arriving on time. It is promising in that, it integrates travel time, risk (associated with the variance in the probability) and deadline. Therefore, this thesis focuses on solving this arriving on time problem, for a single vehicle and multiple cooperative vehicles. Regarding the former, an optimal path for a single vehicle is independently computed using the travel time data of the road links. Regarding the latter, the optimal paths for multiple vehicles are cooperatively computed by exploring their intentions. To circumvent the unrealistic assumptions in the current solutions to the arriving on time problem for a single vehicle, such as Gaussian distribution of travel time, independence among travel time on different road links, and relatively large deadlines, a data-driven approach is developed. More specifically, the PT model based SSP problem is first reformulated as a cardinality minimization problem by directly utilizing travel time data on each road link. Then, this minimization problem is approximately solved via relaxing the cardinality by L1-norm and its variants, and further formulating it as a mixed integer linear programming (MILP) problem. Consequently, this arriving on time problem becomes solvable via an existing MILP solver, e.g., branch and bound (B&B) method. To improve the computation efficiency of the data-driven approach, the property of total unimodularity (TU) that exists in the equality constraint of the MILP-based arriving on time problem is further explored. This nice property can guarantee integer solutions by solving the corresponding linear programming (LP) problem if there is no inequality constraint. In view of this, the partial Lagrange multiplier method is employed to relax the undesired inequality constraints in the MILP problem by shifting them to the objective function. After that, this relaxed problem is solved using the subgradient method in an iterative manner, and only LP problems need to be solved in each iteration, which saves computation cost significantly. To increase the accuracy of finding the real optimal path (i.e., considering that $\ell_{1}$-norm relaxation is an approximation to the cardinality minimization), a practical Q-learning method is developed. In particular, this Q-learning method aims to maximize the expected reward regarding whether the vehicle can arrive on time or not on a specified path. At the same time, the expected reward has the meaning of probability of arriving on time from a current location to destination. To address the issues of continuous deadlines and large-scale road networks, a practical value function approximation method is further proposed. Specifically, it adopts dynamic neural network to directly learn the optimal Q-value in each iteration, which which represents the probability of arriving at the destination on time. %Considering that the optimal Q-value represents the probability of reaching destination from current location before corresponding deadline, hence, results in more accurate solution. %to the PT model based SSP problem. To address the arriving on time problem for multiple cooperative vehicles, a decentralized multiagent-based approach is proposed. It aims to increase the chances of arriving on time for all relevant vehicles. In this approach, two types of agents are involved, i.e., vehicle agent and infrastructure agent. Each vehicle agent follows the route guidance by the local infrastructure agent, and the infrastructure agent collects intentions (i.e., destinations and preferred deadlines) from local vehicle agents. The infrastructure agent formulates the guidance as a route assignment problem, the objective of which is to minimize the cardinality of any possible delays for all local vehicle agents. At the same time, the route assignment problem for each infrastructure agent is solved by existing solvers. Additionally, the proposed route guidance approach is also improved in the aspects of travel time prediction for the assigned road link, computational efficiency of solving route assignment, and acquirement of real-time traffic condition. Extensive experiments on both artificial and real road networks are conducted to evaluate all the above proposed methods. The experimental results confirm the desirable advantages of the proposed methods over others. Therefore, the research in this thesis provides important contributions to the development of intelligent transportation systems. Doctor of Philosophy (IGS)

An efficient power conditioning system plays a significant role in the design of battery charging systems for plug-in hybrid electric vehicles (PHEVs). A typical PHEV battery charging system consists of two stages. The first stage is an ac-dc conversion stage which regulates the input power factor, input current total harmonic distortion and intermediate dc bus voltage. The second stage is a dc-dc conversion stage which regulates the output voltage and provides galvanic isolation between utility grid and PHEV battery pack. The research works presented in this thesis focus on the dc-dc conversion stage of a PHEV charger. The phase-shift modulated (PSM) isolated full-bridge (FB) dc-dc converter topology is commonly preferred for the dc-dc stage. High efficiency, high power density, high reliability, and galvanic isolation are the main features of this topology. However, zero-voltage switching (ZVS) is not ensured for all switches at light-load conditions resulting in poor efficiency of the converter. Furthermore, high voltage spikes are present across output rectifier diodes due to voltage ringing on the secondary side. These voltage spikes are further intensified with the increase in series inductance or leakage inductance of the transformer, output filter inductance and switching frequency. In addition, the voltage spikes increase the electromagnetic interference (EMI) of the converter. Therefore, the motivation for the research work presented in this thesis is to address these drawbacks in order to obtain an improved performance for the dc-dc converter over the entire operating range. Auxiliary circuits can be added to the FB converter for improving the ZVS range. Additionally, snubber circuits can minimize the voltage spikes. In this research work, two new gating techniques, namely asymmetrical pulse-width modulation (APWM) and trailing-edge pulse-width modulation (TEPWM) gating techniques and a passive auxiliary circuit assisted topology are proposed to improve the performance of the dc-dc conversion stage in a PHEV charger. APWM/TEPWM gating techniques minimize the auxiliary inductance value by a factor close to 2 compared to PSM for zero-voltage and zero-current switching (ZVZCS) FB dc-dc converter with auxiliary inductor at both legs. An adaptive frequency control, namely asymmetrical duty cycle frequency control (ADFC) method is also proposed to further reduce the auxiliary inductance and improve the efficiency for the same converter. Another research work presented in this thesis comprising a FB dc-dc converter with inductive and capacitive output filter topologies assisted by an auxiliary transformer and auxiliary inductor is proposed. The proposed converter topologies achieve ZVS for all switches over the entire battery charging range. The proposed converter topology with inductive output filter is employed with a diode clamping network on the primary side to minimize voltage spikes across output rectifier diodes. The proposed converter topology with capacitive output filter is operated as a current-driven rectifier to eliminate voltage spikes. These converter topologies can achieve higher efficiency with the proposed APWM technique compared to the conventional PSM technique. In this thesis, the operation and steady-state analysis of each proposed converter topology are explained in detail. The design considerations for the circuit parameters are also given. The improved performance of the proposed gating techniques and converter topologies is validated with experimental results obtained from 100 kHz, 1.2 kW converter hardware prototypes. Doctor of Philosophy (IGS)

In this study, nature-inspired enhanced microscale heat transfer in macro geometry has been achieved. An annular microchannel of mean channel gap 300 µm and length 30 mm is formed by securing a cylindrical insert of mean diameter 19.4 mm within a cylindrical pipe of internal diameter 20 mm. The nature-inspired Inverted Fish Scale, Fish Scale and Durian enhancement profiles are introduced on the insert surface to improve the convective heat transfer coefficient, for a constant heat transfer area of 18.85 cm2. About 600 steady-state measurements have been collected using 22 microchannel profiles, up to 27 flow conditions (350≲Re≲4,600) for each profile using liquid-phase water, and 2 wall heat fluxes of 13.3 and 53.0 W/cm2. Working correlations have been successfully developed for the average Nusselt number and friction factor, to be used in the design of macroscale heat exchangers employing conventional fabrication techniques and yet exhibiting microchannel heat transfer capabilities. Doctor of Philosophy (IGS)

Ventilated roof receives great interest in recent decades due to its effectiveness in reducing the space cooling load and improving the indoor thermal comfort in summer. A common ventilated roof consists of two parallel solid slabs and an open-ended air layer between the upper and lower slabs. Due to buoyancy or mechanical force, the ventilated air carries out part of the heat accumulated in the roofing system and contributes to the reduction of heat flux transmitted into the building interior. Therefore, the present research work aims to analyze and optimize the thermal performance of ventilated roofs in tropical climate, and minimize the heat flux transferred across the roofs with different inclination angles and geometries, materials as well as ventilation modes. However, the heat transfer mechanisms across the ventilated roof from outdoors to the building interior are complicated, which include the conductive heat transfer in the upper and lower roof slabs, the convective and radiative heat transfers in the ventilated cavity, as well as the radiative and convective heat transfers between roof and the outdoor or indoor environment. Therefore, an accurate prediction of the transferred heat flux requires a complete thermo-fluid analysis of the convective airflow in the ventilated cavity, and a fundamental understanding of the thermophysial properties of roofing materials as well as the convective and radiative heat transfer coefficients at the exterior and interior roof surfaces.As the first achievement made in this thesis, a field experiment and modeling analysis are performed for the thermal performance of a flat roof subjected to the tropical climate in Singapore. A field experiment is carried out to examine the impacts of the solar-reflective cool paint on the surface temperature, transferred heat flux and indoor thermal comfort. Based on the experimentally measured data, a correlation is proposed for the coupled convective and radiative heat transfer between the exterior roof surface and the surrounding environment. In addition, the transient surface temperature of the flat ventilated roof subjected to varying indoor and outdoor conditions is modeled via the complex fast Fourier transform (CFFT) technique. Good agreements are obtained between the theoretically modeled and experimentally measured hourly roof temperatures during both sunny and rainy time. The validated model is extended to study the impacts of the solar reflectance of exterior roof surface and roof insulation on the thermal performance of flat unventilated and ventilated roofs. The second achievement of the present thesis is the development of a novel model for the fast and accurate estimation of the heat flux transferred across the naturally ventilated inclined roof. The developed model takes into account the conductive, radiative and convective heat transfers from the outdoors to the building interior. A weighted-average temperature is derived to represent the temperatures of the sol-air, outdoor air, and indoor air through the circuit transformation theory and thermal network analysis. Moreover, correlations are proposed for the convective thermal resistances at the upper and lower cavity surfaces, based on the two- dimensional(2-D) computational fluid dynamics (CFD) analysis of the turbulent natural convection flow in the roof cavity. The transferred heat flux predicted by the developed model with the proposed correlations is compared with that predicted by a full CFD model, and a good agreement is obtained between the two predictions. The present third achievement is the theoretical and experimental analysis of the impact of the cavity width on the natural convective heat transfer in the inclined roof cavities with low width-to-spacing ratios. A three-dimensional (3-D) CFD model is built up firstly to analyze the airflow movement and natural convective heat transfer in the inclined roof cavity. The 3-D model is validated by the experiments performed on the inclined and vertical cavities with low width-to-spacing ratios. A close agreement is found between the theoretically simulated and experimentally measured cavity surface temperatures, airflow temperatures and velocities. After validation, the 3-D CFD model is employed to analyze the impacts of cavity width on the induced airflow velocity and convective heat transfer coefficient in the cavities. The last achievement is the theoretical and experimental study of the thermal performance of forced-ventilated roofs. A CFD model is built up to analyze the airflow movement and mixed convective heat transfer in the forced-ventilated roof. The CFD model is validated further by the experiments performed on vertical and inclined cavities. The validated CFD model is then employed to analyze the impacts of the induced airflow velocity on the mixed convective heat transfer in the forced- ventilated roof cavities. Doctor of Philosophy (MAE)

Correction for 'High-performance one-dimensional halide perovskite crossbar memristors and synapses for neuromorphic computing' by Sujaya Kumar Vishwanath et al. , Mater. Horiz. , 2024, 11 , 2643-2656, https://doi.org/10.1039/D3MH02055J.