Infrastructure Engineering - Theses
Permanent URI for this collection
Search Results
1 - 10 of 57
-
ItemAn Interface-Aware Multiphase Flow and Transport Scheme for Heterogeneous Porous Media With Material Discontinuities( 2021)Material discontinuities in geologic porous media have a significant impact on multiphase flow, resulting in a wide range of highly nonlinear flow responses ranging from capillary blockage to spontaneous imbibition. This emergent behaviour challenges the validity of standard continuum numerical methods relying on the assumptions of continuity of phase saturation and pressure, slow capillary-limit transport, and steady flow. On the continuum scale, the presence and influence of geologic features are observed from the centimetre to the kilometre scale, which is where geological modelling often focuses due to computing constraints. These effects must be incorporated into models through upscaling. The main objectives of this PhD research are to develop a new numerical scheme that properly accounts for the impact of material discontinuities on immiscible two-phase fluid flow and to upscale them for the purpose of field-scale simulation. The new numerical method extends the finite element - finite volume method by incorporating the extended interface conditions into the models. Thanks to an explicit specific discretisation of these discontinuities, enough degrees of freedom become available to honour both saturation and pressure discontinuities that arise at material interfaces from the interplay of viscous, capillary and gravitational forces. The new scheme is verified with semi-analytical solutions and validated with physical experiments. Then it is applied to study the movement of carbon dioxide and dense non-aqueous phase liquid in two- and three-dimensional models of layered rock. A further extension of this method has been developed to simulate rainwater infiltration through fractured fracture rock in the vadose zone. It uses lower-dimensional fracture representations that significantly simplify model construction while taking into account the material discontinuities effect at fracture walls. The method was validated with a physical experiment. The simulation results highlight the influence of material property contrasts, infiltration rate and initial saturation levels on the speed of wetting front in the fracture slots. Finally, an upscaling technique has been developed to obtain the equivalent relative permeability of heterogeneous centimetre to metre scale models. It uses the above numerical scheme to simulate two-phase flow through a core sample consisting of multiple sandstone layers. In contrast to steady-state upscaling techniques, this method does not assume a capillary equilibrium state in a domain. The results indicate that capillary entrapment at material interfaces can promote anisotropy in relative permeability tensors.
-
ItemWalk-sharing - A smarter way to improve pedestrian safety and safety perception in urban spaces( 2021)Although walking has been proven to be beneficial for the physical and mental health of human beings, and has been shown to be a key to sustainable and liveable communities, its modal share has been gradually diminishing with rapid motorisation and urbanisation across the globe. Moreover, challenging walking environments discourage people from walking, especially in the case of walking for transport. Fear of crime has been cited as the most important barrier for which walking becomes unattractive at critical times of the day, even though walking might be convenient otherwise. Fear of crime influences people's choice of route and travel mode. It forces them to avail costlier alternatives, such as taking viable detours, or abandoning walking altogether and switching to alternative, usually motorised, modes of transport. Fear of crime reduces the overall walkability of an urban area, reduces the time spent on walking, and thereby disrupts the benefits that are offered by walking. Traditional approaches aimed at reducing fear of crime in outdoor spaces, comprising of urban design improvements and infrastructural overhauls, are expensive, localised, involve significant time and human effort. Other, more recent, location and IT based approaches, involving safe route recommendation systems, suffer from heavy dependency on crime and other proxy data sources, and have been known to segregate communities by profiling socio-economic groups To overcome the challenges of the existing methods, this thesis introduces walk-sharing. Walk-sharing is a novel form of shared mobility, which is aims to encourage people to choose walking over alternative modes, when it is viable. As people feel safer walking with a companion as compared to walking alone, walk-sharing matches people with similar spatio-temporal interests who are willing to walk to their respective destinations. By ensuring companionship for pedestrians for a part or the entirety of their journey, walk-sharing improves active natural vigilance, thereby reducing their fear of crime. By reducing fear of crime while walking, walk-sharing has the potential to make walking more attractive, thereby improving its modal share for short-distance trips, and consequently, reduce motorised traffic, thereby reducing emissions and congestion. This thesis discusses the fundamentals of walk-sharing, delineates its similarities and distinctions with existing shared forms of mobility, and proposes a conceptual model that is an abstract representation of a possible walk-sharing system. Based on the logic of the conceptual model, this thesis presents an agent-based simulation model to objectively measure the performance of walk-sharing under plausible scenarios. Using theoretical simulations, this thesis presents the sensitivity behaviour of the walk-sharing model, which also shows the logical efficacy of the model itself. Based on justified assumptions on human preferences, this thesis presents a simulation of walk-sharing on a university campus scenario, achieving up to 80% effectiveness in terms of safety improvement. To gain knowledge about the actual preferences of the community about walk-sharing, this thesis presents a survey and its findings, which depict the spatio-temporal preferences, social preferences, and the overall likelihood of people towards participating in walk-sharing. This thesis finally presents a more sophisticated and grounded simulation of walk-sharing, calibrated using information about actual human preferences on walk-sharing from the survey. Results show that walk-sharing is up to 60% effective in terms of safety improvement, while exhibiting spatio-temporal costs that are within the preferable standards of the community. Walk-sharing overcomes the drawbacks of the existing fear of crime reduction approaches by being proactive (independent of crime and proxy crime data), inexpensive (no requirement of major infrastructural modification or significant human effort), and is scaleable and transferable (can be applied anywhere, and can be easily accessed by the community given the abundance of smartphones). In an age of ubiquitous computing, internet of things, efficient location-based services and smartphones, walk-sharing could be the 'smart' solution that promotes walking as a safer mobility choice for spatio-temporally convenient trips, and consequently progress towards more sustainable urban living, by increasing active mobility and reducing motorised traffic.
-
ItemThe role of proactive rerouting in wayfinding and navigation: A path planning approach( 2021)Human wayfinding has been shown to be adaptive to the information perceived during navigation. Consequently, people may deviate from their originally chosen routes for a number of reasons such as avoiding congestion, missing a turn, choosing a more scenic path, or avoiding complicated intersections. The common operation when changing a path is called rerouting and is generally a path recomputation carried out reactively by a route guidance device. Computing reroutes in an ad hoc manner may lead to unsatisfactory scenarios such as facing an overly long detour or the inexistence of alternative paths at all. Commonly used paths -- e.g., the shortest path -- are oblivious to the properties of reroutes along their way. The overall goal of this thesis is to view rerouting as a proactive task rather than a reactive one, and to understand the benefits of viewing rerouting preemptively to build appropriate path planning methods. This thesis presents definitions and methods to render rerouting useful within route guidance systems, and gives two concrete implementations of path planning methods using this strategy. This thesis first lays the groundwork to use rerouting within computational methods. This first contribution develops a rerouting framework comprising definitions and computational implementations to obtain sets of reroutes efficiently. Using this framework, we make an initial assessment of reroute properties and how to use them to aid in urban navigation. Having laid the foundations to use rerouting in path planning, two specific implementations are presented that capitalise on using reroute properties. These two path planning methods each constitutes a contribution in this thesis. The first method -- called the most recoverable path -- is a path that minimises the impact that navigation errors may have on travel time. Such a path is useful for people navigating an unknown environment or that lack orientation skills. The most recoverable path examines recovery reroutes -- i.e., the shortest path to the destination after a navigation error is made -- and determines their length and the probability of making that navigation error. The second method is called the most flexible path. This path maximises the number of `good quality' reroutes along its way. These reroutes can then be used in multiple ways including overcoming congestion or avoiding complicated intersections. Algorithms to implement these two path planning methods are computationally prohibitive so we introduce efficient approximations to the problems. Additionally, we assess both methods in agent-based simulations in urban settings. These two methods showcase how proactively using rerouting can be beneficial to support dynamic navigation, which is prevalent in urban wayfinding. Overall, this thesis takes a fundamental and powerful idea -- to use rerouting proactively at the path planning stage -- and showcases the benefits of this new paradigm. The algorithms presented here are readily deployable and, therefore, contribute to route guidance systems immediately. Also, future research can extend the concepts and methods presented here to continue building a collection of algorithms that support dynamic human navigation.
-
ItemModelling of Flexible Bodies with OpenFoam( 2021)Over the last six decades, the interaction of floating solid bodies with water waves has received increasing attention from researchers and engineers. Water waves and floating solid bodies can mutually influence each other, depending on the physics of problem. To efficiently design marine structures and accurately predict the wave propagation pattern at covered seas, engineers and scientists require a thorough understanding of the wave-structure interaction. A floating solid body is likely to attenuate the energy of waves and alter the phase speed at which waves propagate. Water waves, on the other hand, can induce motions to and displace the solid object across the sea domain. The wave-body interaction primarily depends on the mechanical behaviour of the floating object (e.g., rigid, elastic, or viscoelastic condition). If the solid body is very thin, the interaction is most likely to be nonlinear, and the body can flex under the force generated by water waves. As such, shear stresses may emerge and dampen the wave energy, and solid body may experience nonlinear motions. Consequently, theoretical models, which are often developed on the basis of potential flow and linear elastic theories, may not be able to reliably predict the outcomes of the wave-body interaction problems. To address this limitation, the fluid-solid body problems should be solved by coupling Computational Fluid Dynamics (CFD) and Computational Solid Dynamic (CSD) models and considering viscous fluids with nonlinear solid motions. In the present thesis, numerical simulations are performed to analyse the interaction of water waves with thin floating solid bodies using coupled CFD and CSD models of OpenFoam, which is an open source numerical toolbox containing various developed solvers for continuum mechanics problems. The primary aim is to determine the ability of the numerical models in simulating the wave-structure interaction problems, and to gain insights into the energy dissipation and dispersion processes when waves interact with different floating solid bodies. To this end, solid bodies with three different mechanical behaviours including rigid, elastic, and viscoelastic conditions are considered to cover a wide range of real objects exist in coastal seas and the ocean. The rigid body simulations are performed by using a CFD model of OpenFoam, which is coupled with a rigid body dynamic solver. The mesh motion is modeled by using an overset technique, which enables us to simulate large motions. The model is seen to predict attenuated waves with a reliable level of accuracy. The energy damping occurring above and under the floating body is found by sampling the fluid field around the body. It is observed that the energy damping emerging above the thin floating body increases with an increase in the wave steepness, and can be formulated by a simple dam-break theory. Similarly, the energy damping occurring under the floating body is observed to increase with an increase in the wave steepness. It is found that the energy damping under the floating body can be estimated by using the work rate of the frictional force. The elastic body simulations are run by coupling CFD and CSD models of OpenFoam, the latter which solves the nonlinear solid motions by using a Finite Volume Method (FVM). Fluid and solid motions are coupled in a two-way framework. The coupled Fluid-Solid Interaction (FSI) model of OpenFoam is seen to predict the wave dispersion process under the elastic bodies with an acceptable level of accuracy. Throughout numerical simulations, it is demonstrated that the elastic body can increase the length of water waves, mostly when incoming waves are relatively short. A body with a larger elasticity is seen to affect the dispersion more significantly. In addition, the FSI model is applied to understand whether the model obeys the available scaling law. By running the model for different scales, it is shown that the FSI solver well follows the scaling law. Simulations are also run for a large-scale condition, which is then paired with small-scale simulations to establish an equation, which rapidly calculates the changes in dispersion that the model gives for an elastic body. The results of this equation are seen to match with the field data. Finally, the interactions of floating viscoelastic bodies with water waves are numerically reproduced by coupling CFD and CSD models of OpenFoam. The computational solid model solves the linear solid motion of a viscoelastic material by employing Maxwell theory. The computational fluid-solid solver is found to compute the energy attenuation and changes in the dispersion process with an acceptable level of accuracy. The phase speed of the wave under the body is computed at the different locations, which is seen to locally vary. But the variation of phase speed from the average value is observed to be subtle. Furthermore, the model is seen to obey an available scaling law. The application of the model in simulating the wave-ice interaction is evaluated by numerically replicating the previous flume and field experiments. The viscosity of the ice is found by using manual fitting. The attenuation rate of the ice is seen to be sensitive to viscosity and is observed to reach a maximum value, which is consistent with the mathematical formulation behind the Maxwell model. The viscosity that gives more accurate results for freshwater ice is seen to be smaller than the viscosity that gives the best fitting for sea ice. This discrepancy is likely to be related to the difference between the natures of freshwater ice and sea ice, artificial damping, and scaling effects. A wider calibration is recommended to be performed in the future to simulate the wave-ice interaction more reliably. Overall, the results of the present thesis confirm that the coupled CFD-CSD models of OpenFoam can help us to model the interaction of water waves with thin floating bodies, especially when nonlinear phenomena, such as shear stresses and nonlinear solid motions, are involved. In addition, the role of the mechanical behaviour of the material is well understood throughout stepwise simulations performed in the present thesis. To sum, the nonlinear wave energy damping is observed to occur above and under a rigid body. For bodies that are likely to flex under the force of waves, changes in dispersion occur. If the body shows an elastic behaviour, the solid-induced energy damping does not emerge. But, when the solid body is viscoelastic, solid motions lead to energy damping, which is not well predicted by an available theoretical model. The findings of the present research can have application in different fields of engineering and science. Firstly, the results and the setup presented for the rigid body can be used in the design process of breakwaters and energy converters, where the solid body is more likely to be rigid and computation of the wave energy is very important. In addition, observations and the setup presented for elastic and viscoelastic bodies interacting with water waves can be used in the design of Very Large Floating Structures (VLFS) and modelling of the wave propagation in covered seas. It is recommended that the model setup is further used to compute the equivalent stresses occurring in the solid body, which can be very helpful in the structural design of floating bodies, and in prediction of the breaking of ice sheets interacting with waves.
-
ItemBehaviour of Driven Battered Minipiles: Field, Physical and Numerical Modelling under Lateral Loading( 2021)The driven battered minipiles have emerged as a sustainable alternative to existing footing systems due to their low carbon footprint and ability to carry complex loads. While the physical attributes of the minipiles are smaller than conventional piles, the applications of the former are similar to that of micropile footings. The driven battered minipiles can be installed rapidly and in inaccessible areas to retrofit the foundation of existing buildings or as new footings for low to medium weight structures. The practical application of these battered minipiles in the construction industry depends on the development of their reliable design methods under complex loading. However, very limited study has been focused on analysing their performance capability, especially, under lateral loading conditions. As a result, this study focuses on the lateral loading behaviour of driven battered minipiles under static and cyclic loading in various soil conditions. This thesis presents experimental and numerical research to develop a robust understanding and design methods for battered minipiles. The testing regime involved lateral load tests on scaled-down physical models from full-scale minipile at 1g scale, full-scale field investigation and numerical modelling. The first part of the test program involved investigating model pile behaviours under static lateral load on single and battered minipile groups in cohesionless soil. The experimental work was further extended to multi-amplitude and long-term single amplitude cyclic loading stages to capture the evolution of stiffness and hysteresis loops for battered minipiles with an increasing number of low-frequency cycles at a constant loading rate. The second part involved static full-scale lateral load testing of single and battered minipile groups in cohesive soil. The gap between the physical model and prototype testing in two soil types was bridged by accounting for stress scale effects for the 1g scale tests and normalising the load-displacement behaviour with respect to the pile and soil properties. The experimental work in both parts was simulated numerically and the validated model was implemented as a tool to perform parametric studies. The key findings from the static lateral load tests were the force-displacement response and strain profile for the minipiles battered at various angles (0, 25, and 45 degrees) with respect to the vertical in both cohesionless and cohesive soil. The numerical model further synthesised the axial force distribution, bending moment and horizontal soil pressure for the single battered minipiles in both soil types. As a congregated outcome, a semi-empirical analytical method was propounded to estimate lateral load capacities of battered minipiles for a range of pile and soil properties. The cyclic loading stage imparted the perception of the development of secant stiffness and hysteresis loop with increasing batter angle (0, 25, and 45 degrees). As the batter angle increased, the hysteresis loop area increased in both the multi-amplitude and long-term single amplitude loading stages. However, the highest secant stiffness was recorded for the 25 degrees battered minipile for both single and multi-amplitude loading scenarios. A multi-surface hardening constitutive macro element model was adopted from literature and modified to capture the damping response of the battered minipiles. The progression of minipile deformation with the increasing number of cycles was characterised through strain profile and perceived to have no significant variation for steel minipiles. In addition to existing conventional battered pile groups, new orientations of minipile groups were introduced in this study that offered higher lateral resistance. The group interaction through experimental and numerical work discerned shadowing effect for similar battered minipile groups in the sand but not in clay. Finally, the parametric study concluded that the optimum angle of batter under static lateral load is a function of pile and soil property and cannot be accurately estimated from existing empirical solutions.
-
ItemNonlinear acoustic metamaterial for attenuation of low-frequency structure-borne sound in multi-storey timber buildings( 2021)Construction of multi-storey timber buildings up to eight storeys in height has seen significant growth in recent years due to advancements in timber engineering and associated revisions in construction codes around the world. Despite achieving high standards of acoustic insulation for mid to high frequencies, effective attenuation of frequencies in the 20 – 120 Hz range has remained a longstanding unresolved challenge for multi-storey timber buildings. Propagation of unwanted noise into neighbouring rooms occurs mainly in the form of structure-borne vibrations that ultimately radiate sound into the air. To solve this issue, a successful approach would need to rapidly attenuate broadband low-frequency vibrations in timber building structures without interfering with existing structural capacity or adding large amounts of additional mass. It would also need to be seamlessly integrated within the restricted confines of practical structures. To tackle this problem, floors are identified as the most significant source of unwanted low-frequency noise due to footfall impacts. Therefore, while the principles underlying the studied designs apply to walls, ceilings and other elements in timber structures, attenuation of impact-induced vibration in floors is identified as the most promising target for improving low-frequency sound insulation in timber buildings. Born out of a thoroughgoing characterisation of the problem and analysis of the limitations of existing methods, this research investigates the potential for a highly nonlinear vibro-impact-based acoustic metamaterial to attenuate 20 – 120 Hz structure-borne impact sound in lightweight timber floors. The proposed system is investigated experimentally and numerically using a developed and validated hybrid finite element (FE)/rigid body dynamic model, which is shown to be capable of efficiently simulating large metamaterial structures while accounting for the effects of detailed internal resonator parameters. The proposed nonlinear vibro-impact metamaterial system is found to be capable of rapidly attenuating impact-induced structure-borne sound in the frequency range of interest. This is demonstrated in a small-scale experimental prototype beam and a full-scale numerical model of a timber floor structure. The system adds only a relatively small amount of mass to the timber floor structure (i.e., the host structure) and does not interfere with the structural performance (e.g., strength or stiffness) of the host structure. The results also show that alternative geometric configurations will be necessary if the system is to fit within the confines of a real timber floor. Through parametric modelling using the validated FE model, parameters contributing to increased nonlinearity in the vibration response are found to improve attenuation. These results demonstrate a potential new approach for dealing with a longstanding issue in multi-storey timber buildings. This is a significant step towards the development and implementation of the proposed solution that would bring about a necessary and significant improvement in building occupant comfort. The results show that it is possible to overcome the limitations of existing approaches using highly nonlinear local resonance effects. Further research is recommended to assess the potential for alternative geometric configurations to achieve similar results in a practical form factor that will fit within the confines of a real timber floor structure. To achieve the development of a practical solution, this work will need to be extended to consider internal structural effects and resonator placement. The capacity of this type of system to attenuate broadband low-frequency vibration opens the door for a wide range of possible designs capable of attenuating unwanted broadband vibration in building structures (including those caused by footfall impacts, vibrating machinery, wind, and earthquakes), bridges, pipelines, and aircraft.
-
ItemNo Preview AvailableOn The Estimation of Wave Loads on Offshore Structures Using Phase Resolving CFD Methods( 2021)With the depletion of the relatively easily accessed resources inland and nearshore, industries have been marching forward into deeper waters. Sea waves are one of the fundamental environmental loads in the design and operation of offshore structures. Consequently, wave loads strongly impact both the design and operational costs of offshore investments. Moreover, it also influences the safety of operators and machinery aboard. For this sake, this research project has been dedicated to investigating two main aspects of wave-induced loads on offshore structures by means of numerical CFD modeling. The first aspect is to evaluate and optimize the implemented CFD numerical model to increase computational efficiency and reduce numerical cost. The second is to investigate potential hydrodynamic approaches to mitigate or reduce hydrodynamic loading of sea waves on offshore structures. In the present study, first, water wave generation by means of imposing a time-dependent Dirichlet velocity profile in a wave flume is investigated. For this sake, the classical wavemaker theory is revisited and adapted for the theoretical formulation of the problem at hand. The validity of the proposed theoretical models is investigated utilizing numerical CFD simulations of two-dimensional wave tanks for different wave conditions. Furthermore, implementation of the proposed wavemaker designs in a typical engineering application is investigated using three-dimensional numerical CFD simulations and experimental tests to validate the outcomes of the numerical models. Second, the derived theoretical formulae are implemented in extending the range of applicability of the conventional active wave absorption method outside the shallow-water waves regime in numerical CFD models. This is done by geometrical modification of the conventional shape of the classical piston wavemaker by a limited absorption depth to reduce the classical mismatch between the absorbing velocity profile and the actual particle kinematic velocity profile of an incident wave. A theoretical analysis is derived to correlate the optimum value of the limited absorption depth to the incident wave condition. The proposed theoretical analysis is validated by means of numerical CFD simulations. Moreover, implementation in a typical engineering application of wave forcing on a vertical rigid cylinder was investigated and validated against available experimental observations. Finally, several hydrodynamic approaches to reduce hydrodynamic induced inline and transverse forces on circular structures subject to water waves are investigated numerically and experimentally. First, the influence of unsteady base injection for a vertical rigid cylinder on hydrodynamic induced loads is investigated for an inertia-dominated flow. Different injection orientations are investigated entailing kinematic profiles, amplitudes, and synchronization angles. Second, the influence of steady injection around a circular cylinder in a viscous unsteady flow is investigated as a hydrodynamic means of reducing vortex-induced vibrations (VIV). The hydrodynamic inline and transverse forces are investigated for different angles of attack and injection amplitudes.
-
ItemPost-processing Sub-seasonal to Seasonal Climate Forecasts Under Climate Change( 2021)For managing the impacts of climate variability and change, climate outlooks on sub-seasonal and seasonal timescales are attracting more interest from climate-sensitive communities, such as water resource management, agriculture, and energy. With a profound knowledge of the sources of climate predictability, modelling techniques are rapidly developed for forecasting future climate conditions. Recent advancements are dynamical global climate models (GCMs), which typically integrate atmosphere, land surface, ocean, and sea ice components to comprehensively simulate earth climate system and output a wide array of climate forecasts. However, GCMs usually suffer from long-standing modelling issues, such as systematic errors and the failure of reproducing the observed trends in seasonal climate forecasts. Statistical post-processing techniques are frequently employed to improve forecast performance. Many commonly used methods are found to be effective at removing biases, maximising forecast skill, and improving forecast reliability in terms of ensemble spread, but they are seldom designed to resolve the trend disparity issue in the post-processed climate forecasts. This issue should not be neglected as global and regional land surface temperature and precipitation have shown discernible temporal trends over recent decades. To address this gap, the overarching objective of this thesis is to develop and demonstrate the merit of a new, trend-aware forecast post-processing method that eliminates the trend disparity between climate forecasts and observations while making the forecasts bias-free, skillful, and reliable. The first part of this research aims to develop a new statistical post-processing method to embed the observed trend into seasonal temperature forecasts. I extend the capability of a calibration method, the Bayesian joint probability (BJP) modelling approach, by introducing a new trend component into the algorithm. The new model (named BJP-t) is applied to calibrate January mean forecasts of daily maximum temperatures from the SEAS5 seasonal forecasting system, operated by the European Centre for Medium-Range Weather Forecasts (ECMWF) in three test stations in Australia. In these cases, the BJP-t calibrated forecasts are shown to accurately reproduce the observed trends, and are more skillful, more reliable, and sharper than raw and BJP calibrated forecasts. In the BJP-t model, the trend is entirely inferred from the training data. In practice, given limited available periods of retrospective forecasts for model training, these inferred trends are subject to large sampling errors, and may not reflect true underlying trends in the observations. Accordingly, the second part of my thesis further develops the BJP-t model to account for trend uncertainty. The extended trend-aware forecast post-processing method is applied to SEAS5 seasonal mean minimum and maximum temperature forecasts, and the evaluations are upscaled to the Australian continent. After trend-aware post-processing that deals with trend uncertainty, forecast trends are more robustly inferred than the BJP-t model. Compared to the BJP calibrated forecasts, embedding trends lead to greater forecast accuracy in regions where observed trends are significant or where observed trend direction is wrongly represented in the BJP calibrated forecasts. The third part of my thesis aims to extend the trend-aware method for post-processing seasonal forecasts of precipitation, which is also a key variable for agriculture and water resource management. Several modifications are made in the model algorithm and evaluation tools to cater for the special features of precipitation amounts, including zero occurrences, highly positive skewness, as well as higher variations and larger uncertainty than temperature variables. I apply this improved trend-aware method to calibrate SEAS5 seasonal precipitation forecasts over Australia. Evaluations show that the trend-aware calibrated forecasts properly reproduce observed trends over the hindcast period of 36 years. In some regions with significant observed trends, skill improvements against the BJP calibrated forecasts are evident by embedding trends into the forecasts. Overall, in most regions, the trend-aware calibrated forecasts outperform raw forecasts with respect to bias, skill, and reliability. Operational sub-seasonal climate forecasts are produced by GCMs configured not dissimilar to seasonal forecast models, but little attention has been paid to explore the ability of the sub-seasonal forecasting systems to capture the observed trends. The fourth part of my thesis firstly aims to investigate this question. Preliminary results show that the same trend disparity issue exists in the 20-year weekly averaged retrospective temperature forecasts from the ECMWF extended-range forecasting system, particularly beyond the first week. Subsequently, I adapt the trend-aware method to calibrate and correct the trend in sub-seasonal forecasts. I modify the method to embed a 30-year climate trend into the 20-year calibrated forecasts. The embedded trends are therefore robustly representative of long-term climate changes and overcome the problem that trends inferred from a shorter period may be subject to large sampling variability. Calibration is applied to the ECMWF sub-seasonal minimum and maximum temperature forecasts for Australia with forecast horizons of up to 4 weeks. Results reveal that raw week-1 forecasts exhibit trends consistent with the 20-year observations in many regions while raw week-4 forecasts do not show the trends of the 20-year observations during the hindcast period. After trend-aware post-processing, trends in calibrated week-1 forecasts are roughly aligned with the 20-year observations across Australia, because when raw forecasts are inherently skillful, the trend-aware calibration transfers raw forecast skill and embeds the 20-year apparent observed trends into the calibrated forecasts. For comparison, calibrated week-4 forecasts exhibit the trends of the 30-year observations, because when raw forecasts do not have much skill, the trend-aware calibration reverts the forecasts to the 30-year observed climatology with trends. In general, the trend-aware calibrated forecasts are more reliable than raw forecasts, while being as skillful as or more skillful than raw forecasts for all lead times. The new trend-aware forecast post-processing method shows robustness for resolving the trend disparity issue for GCM sub-seasonal and seasonal climate forecasts. Wider applications of this method have the potential to deliver quality forecasts and build user confidence in deploying the forecasts for decision-making in a changing climate. Further research will adapt the trend-aware method for other hydrometeorological variables.
-
ItemMultispectral and Hyperspectral Remote Sensing of Canopy Nitrogen Concentration( 2021)The concept of nitrogen (N) fertilizer optimization is central to achieving the dual objective of optimizing the crop yield while reducing the environmental impact in precision agriculture. Canopy nitrogen concentration (CNC) is a measure of crop N status that can be instrumental in mapping the crop N requirement in space and time to optimize N fertilizer application. Beyond the cropland, CNC information is vital in monitoring the ecosystem functioning through its linkage with carbon and nitrogen stocks. Remote sensing offers a rapid, non-destructive and cost-effective way of measuring the CNC status of the crops in contrast to the time-consuming, destructive and costly laboratory measurements. Therefore, based on the linkage between CNC and canopy reflectance, many remote sensing models have been proposed for CNC estimation. However, their mapping capability is greatly influenced by confounding factors such as biomass, canopy architecture, and soil background, which vary with growth stages, seasons and crop type, limiting their broader adoption and application. Therefore, this thesis proposes robust CNC remote sensing models that are deployable to cost-effective sensors. There are many multispectral CNC indices, but little is known about their robustness across a wide range of contrasting growth stages and seasons. Chapter 3 assesses the robustness of the eighteen widely used multispectral indices for CNC sensing with an aim of enhancing the current understanding of their performance. The results indicate that almost all the indices exhibit a similar level of correlation with CNC when applied at individual growth stages; however, the multispectral index vs. CNC relationship is growth-specific and varies significantly across the growth stages. Therefore, this unstable relationship leads to significant performance drops when evaluated against the pooled data. It is found that the majority of the CNC indices respond to CNC and biomass together with an even more consistent association with biomass than CNC. Furthermore, some growth stages exhibit a very high correlation (up to 0.90) between observed CNC and biomass which may further obscure the identification and development of robust CNC indices when only limited growth conditions are incorporated. Among the selected indices, only the photochemical reflectance index (PRI) exhibited a consistent, albeit low, association with CNC. Motivated by the limitations identified in Chapter 3, Chapter 4 implements an exhaustive search to optimize the 2-4 waveband based multispectral indices for CNC. This search incorporates a wide range of growth stages in seasons of two crop types, ryegrass (irrigated) and barley (rainfed). The results show that although the best waveband combination changes with season and crop type, the visible spectrum, especially the blue region, exhibited consistent sensitivity to CNC. The newly developed 4-waveband index (ND4) is a more skillful predictor with stronger linearity with CNC than its 2-waveband counterpart (ND2). In addition, ND4 effectively reduces the predictive loss when broader wavebands are used compared with ND2. Furthermore, the issue of biomass and canopy structure influence has been significantly reduced in both ND2 and ND4 compared with the pre-existing indices considered due to the use of a diverse data set in the index development process. Chapter 5 explores the full potential of the information-rich hyperspectral data by employing state-of-the-art statistical, machine learning (ML) and deep learning (DL) techniques to model CNC. Models to predict aboveground biomass (AGB) were also developed to characterize the canopy N status across the growth cycles using the ‘CNC dilution’ effect. These modelling approaches include partial least squares regression (PLSR), random forest (RF), 1D and 2D convolution neural networks (CNN), together with comparison against multispectral indices, ND4 and soil-adjusted vegetation index (SAVI) for CNC and AGB, respectively. The results indicate that 1D CNN is the best and its performance is followed by the PLSR model. These models are more robust and generalizable than their hyperspectral counterpart RF and 2D CNN. The multispectral ND4 index exhibits performance comparable to the PLSR in the test dataset but with lower generalizability. In general, the models underperform in low canopy cover for CNC and under the high biomass for AGB prediction. Furthermore, intercomparison among the spectral regions shows that the visible spectrum is a more informative input for the CNC models while the near-infrared (NIR) spectrum is for AGB models. The thesis provides an enhanced understanding of the robustness of multispectral indices for CNC estimation. The proposed multispectral indices in this thesis only employ 2-4 wavebands and are consistently sensitive to CNC across contrasting growth conditions. In addition, this research also investigates the potential and limitations of the hyperspectral models to estimate CNC and AGB.
-
ItemMultiscale Modelling and Homogenization of Ultra-High Strength Concrete (UHSC) from Nano to Macro Scales( 2021)Ultra-High Strength Concrete (UHSC) is now being widely adopted in the construction industry due to the very high compressive strength, high elastic modulus, excellent durability characteristics etc. In the macroscopic scale, UHSC is considered as a homogenous and isotropic material when designing concrete structures. However, in reality UHSC is a highly heterogeneous material consisting of different length scales of heterogeneities known as multiscales of concrete. These multiscales of concrete can be identified as macroscale, mesoscale, microscale, nanoscale, and atomic scale. Mechanical and durability characteristics and the nonlinear behaviour of UHSC in the macroscale depend on the characteristics of the constituents in these multiscales. Modelling these individual scales and developing a connectivity between the scales are essential in exploring how the phases in each scale are contributing to the macroscopic behaviour of UHSC. Even though previous researchers have focused on individual scales of UHSC, a comprehensive study examining all the multiscales present in UHSC and developing a framework to link these scales to explore the contribution from each scale to the macroscopic behaviour is lacking. This research tries to address the question of how to model UHSC in different scales and how the intrinsic material properties of constituent phases in the different spatial scales contribute to the overall macroscopic behaviour. The main aim of this investigation is to create a comprehensive multiscale modelling framework to model UHSC in various length scales, and to bridge the gap between these scales and upscale and predict the macroscopic behaviour of UHSC using constituent properties at nano, micro and meso scales. In this research, mesoscale, microscale, and nanoscale of concrete are modelled using advanced modelling techniques such as finite element representative element modelling (FE-RVE) and finite element mesoscale modelling. An experimental programme is carried out to investigate the material parameters of the constituents at different scales using macro scale experiments and nanoindentation to characterize the nano and micro properties. The properties from nanoindentation results were used to evaluate the different phases and their homogenized elastic and strength parameters using statistical deconvolution, finite element limit analysis, Linear Comparison Composite (LCC) method, and an inverse analysis algorithm. Proposed framework is capable of homogenizing the elastic and strength parameters and linking the micro and nanoscale properties to the FE-RVE models so that the material properties can be upscaled from nanoscale to the macroscale. Finally, the homogenized mechanical properties are compared with the experimental properties of UHSC at the macroscale and the results concluded that the proposed multiscale and homogenization framework is able to predict the macro properties using nano and microscopic properties. Mesoscale modelling of concrete to explore the fracture and damage propagation behaviour of UHSC investigated in this thesis as the first component. Aggregate scanning and synthetic aggregate generation methodologies using spherical harmonics and other algorithms are presented and the behaviour of UHSC under uniaxial compression is investigated using mesoscale modelling. Experimental procedure followed to characterize UHSC microstructure and the nanoindentation tests performed to obtain the micromechanical properties of UHSC are presented subsequently. Also, the hydration simulation of UHSC is examined and compared with the High Strength Concrete (HSC). Finally, the elastic modulus and compressive strength homogenization for UHSC using the continuum micromechanics based analytical models and numerical FE-RVE modelling is presented. It can be concluded that the modelling of UHSC in different spatial scales can identify the critical phases which contribute to the enhanced mechanical properties in the macroscale and the developed multiscale modelling framework can successfully upscale and predict the mechanical properties of UHSC using the micromechanical properties of constituents in different spatial scales. This will also help to accurately model structures at macro-scale.