The urbanisation and population growth are resulting in a significant increase in energy consumption in buildings, leading to a substantial increase in greenhouse gas (GHG) emissions. During the operation of buildings, a massive amount of GHG emissions are released due to the process of building heating, cooling, and lighting, which accounts for the most significant proportion in building energy consumption. Therefore, energy-efficiency design and operation will play an essential role in reducing GHG emissions in buildings. Facade systems are one of the most critical aspects regarding the efficient management of heating, cooling, and lighting energy in buildings. A facade system is a barrier and exchanger (simultaneously) for temperature, light, and air between the building indoor environment and the outside environment. Therefore, the proper design and operation of the facade can effectively save substantial energy. For decades, engineers and researchers from all over the world have been in search for the intelligent design and operation of the facade systems to improve energy efficiency and sustainability in buildings, and to not compromise a pleasant indoor environment for building occupants. Subsequently, they have found that many natural systems have developed a highly efficient biological structure to adapt to dynamic and extreme environments over millions of years. These natural systems now have become great inspirations for the research community in the quest for building energy efficiency solutions, and the biomimetic adaptive facade (BAF) system is one of those remarkable examples of adopting bioinspiration in buildings. The BAF system is considered as a potential solution to improve the performance of conventional facade systems. The BAF system has an ability to adapt its functions, features, or behaviour for dynamically varying climatic conditions, providing buildings with the operational flexibility to act in response to different climate scenarios. Nonetheless, the practical application of a BAF in buildings remains limited due to the absence of a comprehensive design platform that can facilitate the widespread adoption of BAF systems. Most studies on BAFs remain at a conceptual stage of development, and an effective platform that can effectively assist the design and operation of BAF is still lacking. This thesis proposes and develops a methodology for enhancing building energy efficiency using the design of BAF systems, and thereby supports the transition to next-generation facades. Specifically, the objective of this thesis is to develop, test, and evaluate a computational data-driven optimisation approach in assisting the BAF design. The thesis presents a multidisciplinary approach that combines building energy modelling, metaheuristic optimisation, and data-driven methods. The goal of the proposed approach is to minimise the total energy consumption in buildings, including heating, cooling, and lighting energy, but still maintain the indoor environmental quality in terms of thermal and visual performance. A comprehensive analysis of the proposed computational data-driven optimisation approach is provided in the thesis. In summary, this study has proposed a computational data-driven approach based on building energy simulations, optimisation processes, and machine learning algorithms. The proposed approach is used to assist the design and operation of BAFs for building energy efficiency and analyse the interactions between energy-saving and indoor environmental quality. These significant findings demonstrate the potential of BAFs to enhance the energy efficiency of buildings, and the developed platform can be used as an effective tool to support BAFs in both design and product development.

Concrete flat slabs are widely used, especially in multi-storey buildings, because of their resource efficiency and fast construction. The requirements for fire safety critically affect the overall design of these slabs. Fire design guidelines which are based on research carried out a few decades ago hinder the effective use of concrete flat slabs as they impose strict thickness and cover requirements. Since then, material properties of concrete have significantly changed, and construction methods have evolved considerably. Therefore, new research is needed to assess the fire performance of concrete flat slabs and provide a research base to improve the current fire safety design guidelines for concrete flat slabs. Among the limited number of fire tests performed on concrete flat slabs, most of them were simply supported isolated specimens which did not take into account the continuity of the slab. Therefore, the author conducted a large-scale fire test on a laterally restrained flat slab specimen simulating the effect of adjacent slab panels in case of a fire. Recent studies emphasise the importance of capturing the behaviour during the cooling phase as there is a risk of failure also during that phase. Hence, the experiment was extended to measure the thermal and structural response during the cooling phase. Results indicate that the fire resistance level (FRL) of the restrained flat slab has been improved compared to the FRLs predicted by the design standards. The use of restrained support conditions which allows the development of membrane actions could be the reason for improved FRL. Although fire tests provide vital information on the behaviour of flat slabs in fire, they are very expensive and time-consuming. As an alternative, numerical methods can be utilized to capture the effects of elevated temperatures on concrete flat slabs. Building upon the existing material models for concrete and steel at elevated temperatures, the author introduced the use of explicit coupled-temperature analysis technique in finite element software ABAQUS to determine the thermal and structural response of concrete flat slabs in fire. The models also account for the transient thermal creep of concrete when heated and change of material properties during the cooling phase. The developed models were validated with the experimental results from the fire test carried out by the author, along with two more independent fire tests. A validated numerical modelling technique was then employed in a parametric study to evaluate the influence of the thickness, the span between columns and the reinforcement arrangement on the FRL of concrete flat slabs. The aim of the study was to further understand the new design rules imposed by the latest Australian concrete design code. Outcomes of the analysis further validate some amendments incorporated in the design code while suggesting improvements to the critical distance rule. The current construction industry prefers performance-based fire design over prescription-based fire design as it yields more optimized solutions on a case by case basis. In order to apply such methods to concrete flat slab fire design, a case study was conducted to model an actual fire scenario within a compartment in a multi-storey building and its effects on the flat slab. Fire Dynamics Simulator (FDS), a computational fluid dynamics based fire simulation software was implemented to capture the growth and decay of a compartment fire incorporating the combustible material characteristics of the furniture inside. Different fire scenarios were simulated, taking into account the different ventilation conditions. Critical temperature fields generated from the fire simulation were then applied to a FE model to assess the structural response. The predicted structural response for the actual fire scenario is significantly different from the response when the flat slab was subjected to standard design fire. This observation further highlights the importance of performance-based fire design approach, which takes building-specific parameters into account rather than generalized fire curves.

Tidal inlets are located between open seas and coastal lagoons. They experience complex hydrodynamic patterns owning to the actions of ocean waves and tidal currents - the two main drivers of various coastal processes including sediment transport, shoreline evolution, erosion, inundation and extreme sea levels. Better decision-making, sustainable management and adaptation requires numerical models to predict coastal hydrodynamics to gain a better understanding of how these mechanisms behave and change over time. Especially nowadays, human-induced pressures have been increasingly causing significant disruption to the formerly pristine state of these important coastal systems. However, our existing knowledge and the reliability of existing models are far from adequate. This study aims to provide a better understanding of coastal hydrodynamics, including the impacts of human-induced influences on these fields. The contributions of this research include the implementation of advanced measured-based physical packages (ST6) in a spectral wave model - WWMIII - which provides an important update in terms of predicting water waves focused on coastal regions. This implementation was tested in several water environments before being integrated into a coupled hydrodynamic model system (SCHISM-WWMIII) to study climatology of hydrodynamics at Port Phillip Bay, Australia based on long-term hindcast data. These data were derived from 27-year simulations applied to two bathymetry options: before and after channel deepening at Port Phillip Bay. To this end, these decadal data sets provide a unique opportunity to quantify hydrodynamics and possible impacts due to dredging on these fields in more detail.

The 2030 Agenda for Sustainable Development, anchored by 17 Sustainable Development Goals (SDGs), 169 targets, and a global indicator framework, provides a transformative and integrated approach to sustainable development. Although not readily apparent, the SDGs are highly dependent on geospatial information and enabling technologies as the primary data and tools for relating people to their location, place and environment, and to measure ‘where’ progress is, or is not, being made, particularly at sub-national and local levels. For this reason, the 2030 Agenda specifically references the need to exploit geospatial information and other data sources to ensure that high-quality, timely and reliable data, disaggregated by geographic location and other characteristics, is available – especially to developing countries. However, in the pursuit for sustainable development, many countries continue to face a series of impediments that exacerbate their ability and ‘opportunity’ to participate fully in the implementation of the SDGs, to support national development, economic prosperity, and, through that, a global and thriving information economy. These include institutional challenges in data production: having the required human capital and skillsets; effective and sustained access to digital technology, the Internet and the corresponding computer literacy; to the provision and exploitation of new data needs, information systems, analytics, and associated enabling tools and technologies to support the timely and reliable implementation of the SDGs. They also include policy challenges, including strategic leadership, understanding and awareness of national geospatial information policy, frameworks, and associated implementation roadmaps. Governments are failing in their attempts to address the geospatial needs of the SDGs, as they lack the required guidance to assist them. National data systems are often fragmented, and with no integrative strategic frameworks, roadmaps, and tools in place to determine how geospatial information can be implemented and integrated into the SDGs. These problems are a very real impediment for many developing countries, those most affected by the challenges and need to achieve national development, to being able to fully realise the implementation of the SDGs – and to ensure no one is left behind. As a tangible means to support the implementation of the SDGs, this research investigates the key aspects and impediments as to why geospatial information remains unable to adequately contribute its data, systems, and integrative capabilities to support the measuring, monitoring, and implementation of the SDGs, particularly at a national level. The research addresses this problem through its overall aim; to define, develop, test, and apply a new national strategic geospatial information framework to enable countries to integrate geospatial information into national sustainable development strategies and processes, with particular application to the SDGs. Through the implementation of the strategic framework, countries will be able to measure and monitor progress and transformative change within their national circumstances, and mainstream evidence-based policy-setting and decision-making to achieve sustainable development. To achieve the research aim, a mixed methods research design is employed through the application of archival research and a two-dimensional case study. Archival research is used to: examine and investigate existing concepts on sustainable development theory and practice, and review the relationships with and role of geospatial information; and evaluate the evolution and development of geospatial information, its uptake at the intergovernmental level, applicability to the SDGs, and identify issues and challenges impeding geospatial information from contributing more to sustainable development. Given the emphasis on the SDGs and developing countries, a precursor to the development of the strategic framework was to first review the relevance of the digital divide and the related data availability and integration issues for the SDGs. Therefore, the research introduces and develops the components of the ‘geospatial digital divide’ and the complex challenges that continue to exacerbate the ability for many developing countries to bridge this divide. The research then develops a conceptual integrated geospatial information systems framework as an integrative sustainable development ‘data flow’ to provide the building blocks and processes for countries to measure and monitor the SDGs with relevant and timely location-based data. A two-dimensional case study approach, focusing on developing countries, is then undertaken – regionally within the continent of Africa, and nationally within the country of Ethiopia. The first part of the case study iteratively defines, develops, and tests the prototype strategic framework with experts in African countries. The second part of the case study implements and applies the strategic framework in Ethiopia, through the Ethiopian Geospatial Information Institute. During the course of this research, the legitimacy of the strategic framework was recognised by the United Nations and World Bank as a means to progress an overarching geospatial framework to assist developing countries to bridge the geospatial digital divide. In August 2018, the strategic framework was adopted by UN-GGIM as the Integrated Geospatial Information Framework (IGIF). The outcome of the framework’s adoption by UN-GGIM immediately validated its critical need. The research findings confirmed the need for an overarching strategic framework that could be general enough to address many systemic institutional barriers through aspirational principles and goals, while also being detailed enough to provide the pragmatic pathways and roadmap to overcome the research problem. The IGIF now provides that framework. While linking to, and building upon NSDIs, the strength of the IGIF is that it is not a data infrastructure. It is an integrated framework and a knowledge infrastructure that can be applied to all countries and all situations and circumstances.

Human activities from overpopulation to pollution are drastically increasing the global temperature and causing fundamental changes in the Earth. The primary reason for this phenomenon is the greenhouse gas effect that primarily comes from the destruction of the forest cover and conventional energy production. This would cause extreme weather conditions, consequently imperilling the natural habitats, ecosystems and all the Earth’s activities. Emission of an enormous amount of anthropogenic carbon dioxide (CO2) has become the primary reason for global warming; so, it is vital to advance towards novel techniques to alleviate the issue. Within a portfolio of options, CO2 sequestration has gained growing attention – aimed at carbon capture and storage in geological formations. In this context, CO2 sequestration in coal reservoirs is identified as an effective method, because the process is associated with enhanced coalbed methane (ECBM) extraction – that further offsets the cost of CO2 storage in coal reservoirs. The successful implementation of the process would certainly lead to a pollution-free world and ultimately a better future, through the mitigation of the greenhouse gas effect. Although the fundamental physics of CO2 sequestration have been evolved over several decades, there are some grey areas and unexplored aspects, that need further improvement through investigations. For instance, the injection of CO2 into coal affects its structural properties, and consequently, the efficiency of the sequestration process as well as the long-term integrity of the reservoir. The coal-CO2 interaction is a complicated physico-chemical and thermodynamic process that depends on multiple parameters, which in turn makes the evaluations quite challenging. A deep understanding of the structural alterations, in terms of coal hydro-mechanical properties, is thus essential to optimize the process, while minimizing the reservoir damage. Considering the above aspects, this research aims to comprehensively characterize the alterations in hydraulic and mechanical parameters of CO2-interacted coal, through a combination of systematic experimental, numerical and statistical analyses, integrated with advanced image analysis techniques and chemical analyses. The study is specifically focused on analysing the fully-coupled CO2 flow–adsorption–deformation process in coal and investigating the consequent hydro-mechanical modifications occur in coal structure, due to the complex coal-CO2 interaction. Initially, the fully-coupled process in a single coal matrix block was modelled, in which the multiple diffusion mechanisms in matrix pore network were simulated using a theoretically-extended diffusion modelling approach, and the consequent CO2 adsorption-induced matrix swelling was modelled and validated with experimental results. The analysis was then extended on a fractured coal geometry at the coal constituent-scale, through the inclusion of a 3D – discrete fracture matrix (DFM) network, in which the alterations in local fracture aperture and the subsequent fracture permeability were investigated. In general, the results conclude that the CO2 interaction-induced hydro-mechanical alterations of a fractured coal seam are largely geometry-dependent and highly-localized; thus, should be resolved accurately through a fully-coupled DFM modelling approach. Next, a series of experimental, numerical and statistical studies was conducted to further assess the CO2-interaction-induced complex mechanical property alterations in coal. The alterations in compressive strength, Young’s modulus, brittleness index and dynamic moduli were experimentally investigated, in which the temporal and localized mechanical alterations were specifically analysed in detail. A numerical model was developed to evaluate the free and adsorbed CO2-induced compressive strength alterations in coal, under both confined and unconfined conditions. The compressive strength was determined with the Mohr-Coulomb failure criterion, where a theoretically-modified cohesion model was derived, by integrating the effective stress change and the surface energy change – caused by CO2 interaction. Further, a statistical analysis was conducted using advanced soft computing techniques to predict the mechanical property alterations in coal, under different CO2 interaction conditions. The studies combinedly conclude that CO2 interaction results in adverse mechanical alterations in coal and the developed numerical and statistical models are capable of accurately simulating/capturing this behaviour. Finally, the conclusions, recommendations for future research and an insight into the field-application of CO2 sequestration technique were provided based on the results of the combined analyses. Overall, the outcome of this research is intended to provide knowledge and insight into the productivity and safety of the CO2 sequestration in coal reservoirs, by conclusively evaluating the CO2 interaction-induced structural alterations in coal, in terms of its hydro-mechanical properties. The information, results and conclusions given herein can be utilised by researchers to build upon and will be beneficial on reducing the global warming and climate change, through effectively minimising the atmospheric CO2 level.

The building sector has been at the centre of the environmental protection policies of the European Union (EU). This is mainly due to its high energy consumption (40% of the total) among the EU Members. Residential buildings are responsible for two-thirds of that amount. Recently, the European Commission set radical targets for the reduction of greenhouse gas (GHG) emissions by 2030 (50-55% reduction compared with 1990 levels) and 2050 (climate neutrality) and the building sector, especially the existing building stock, is expected to play a critical role in achieving those goals. Special consideration should be given to multi-residential buildings. Compared to other building types, they have limited suitable space for the installation of renewable energy systems and are governed by a complex legal framework, imposing additional challenges on decision-making. Targeting multi-residential buildings, this study developed a method for the identification of optimal retrofit sets. It is a multi-objective simulation-based optimisation method for the performance assessment of ‘whole building’ retrofit interventions under two objectives: the minimisation of the operating GHG emissions and the life-cycle cost. The innovation in the method is the integrated approach, considering energy supply, energy demand-side technologies and energy-saving measures. A dynamic building systems’ modelling process was also introduced, based on part-load performances, to address the accuracy limitations of existing, monthly quasi-steady state methods. The functionality of the method was illustrated through an application. The case study building is a 6-storey multi-family building, constructed before 1980. The performance of several retrofit sets of measures was compared to the ‘base case’ building, which is a comparable version of the case study building. To identify the way that various parameters of the building environment affect the method application and the obtained results, four locations were considered, one for each Greek climate zone. It was found that for all Greek climate zones, the cost-optimal retrofit set consists of the roof and basement ceiling insulation, the installation of air-to-air heat pumps (HP) for heating and cooling and solar thermal panels for domestic hot water (DHW). This way, the operating GHG emissions could be decreased from 59% to 67%, compared to ‘base case’, depending on the building location. To achieve a retrofit that minimises the GHG emittions (almost 90% less operating GHG emissions compared to ‘base case’), the obtained solution sets included wall, roof, basement ceiling insulation and window replacement with double or triple-glazed windows, central biomass boilers (locations without natural gas) or gas condensing boilers (locations with natural gas) for heating, air-to-air HPs for cooling and photovoltaic-thermal (PV/T) panels for DHW and electricity production. Net-zero carbon retrofit solutions could not be achieved for any location. The findings of the study are in line with the observed market trends (envelope insulation, installation of double-glazed windows and air-to-air HPs, condensing gas boilers or biomass boilers). Gas absorption HPs and air-to-water HPs will be competitive alternatives when their purchase costs decrease. Similarly, solar thermal collectors for DHW are the common practice, however, when solutions that minimise GHG emissions are required, PV/T panels have great potential but limited market penetration. The results obtained are specific to the financial situation, fuel, renewable energy sources and systems availability of the considered locations. They can be used as a guide for retrofitting similar buildings and construction types in urban areas of the Mediterranean climate, assisting policymakers and homeowners.

The present study deals with the displacement based seismic assessment of earth retaining structures. Detailed shaking table experiment has been performed on two different scaled down retaining wall models. The first set of shaking table experiment has been performed on base restrained retaining wall model, and the second set of shaking table experiment has been performed on a rotational base retaining wall model. It was observed from the shaking table experiment that inertial forces from backfill profoundly influence the seismic displacement of the earth retaining structures. Different dynamic properties of scaled down retaining wall models have also been studied with the shaking table experiment results. Significant amplification of horizontal acceleration in the backfill has also been observed during all shaking table experiments. A detailed geotechnical investigation has also been performed on the backfill used for scaled down retaining wall model construction. Characterization of the backfill has been achieved based on different geotechnical experiments. The hardening/softening behaviour of backfill has also been studied from the consolidated drained (CD) triaxial test results and the Mohr Coulomb material model. A detailed process for the finite element (FE) modelling of seismic actions on earth retaining structures has been explained. Capability of the FE models has also been verified by replication of the shaking table experimental results. A detailed and rigorous FE investigation has been performed on different earth retaining structures. The base restrained retaining wall, the free straining retaining wall, and the cantilever retaining wall founded on rock-socketed pile foundation have been considered for the FE investigations. Earthquake induced displacement behaviour of the earth retaining structures has been studied against different synthetic and historical accelerograms. The effects of different backfill type on the seismic response of earth retaining structures has also been studied. Significant amplification of the horizontal acceleration in the backfill has been observed for all cases. Simplified hand calculations have also been proposed for estimating the seismic displacement of the base restrained retaining wall. It was observed that the earthquake induced displacement of earth retaining structures mainly depends on the severity of ground shaking, inertial forces from the backfill, and hardening/softening behaviour of the backfill. In the case of a free-standing retaining wall and a retaining wall founded on rock socketed pile foundation, granular backfill shows better seismic performance than natural sand backfill. In the case of a retaining wall founded on the rock socketed pile foundation, high displacement of the rock-socketed piles has been observed.

The scale-dependency of overland flow is frequently observed in rainfall runoff measurements, (Wilcox et al., 1997, Wilcox et al., 2003, Van de Giesen et al., 2000, Sheridan et al., 2014), yet largely neglected in rainfall-runoff models (Bloschl and Sivapalan, 1995, Chen et al., 2016a). Overland flow scaling behaviours within a given hillslope have been attributed to the main factors controlling infiltration and runoff processes including spatial variability of soil hydraulic properties (Julien and Moglen, 1990), run-on effect (Wainwright and Parsons, 2002, Langhans et al., 2014), macropore flow (Nyman et al., 2010, Ritsema and Dekker, 1995, Wessolek et al., 2009, Nyman et al., 2014, Stoof et al., 2014a, Ritsema et al., 2005), and rainfall temporal properties (Joel et al., 2002, Li and Sivapalan, 2011, Wainwright and Parsons, 2002). These factors are nonlinear and vary in time and space causing uncertainties when averaging between scales. Wildfire may introduce higher spatio-temporal variability to the factors controlling soil infiltration by vast alteration in soil and vegetation, and as a result of that scaling effects on hydrological processes may be altered in burned landscapes (Moody et al., 2013). However, the impact of fire on scaling behaviours is poorly investigated and only few practical studies have measured runoffs scaling on burned hillslopes (Sheridan et al., 2014). There are significant knowledge gaps in understanding overland flow scaling effects in relation to post-fire soil, surface factors and rainfall properties (Moody et al., 2013). This study aimed to investigate overland flow scaling behaviours in relation to soil and rainfall properties on a burnt hillslope by observations, measurements, and simulations. This was obtained by i) collecting rainfall-runoff data from different plot lengths at a eucalyptus hillslope, Southeast Australia burned by wildfire in 2013, ii) quantifying the degree of runoff scale-dependency from empirical rainfall-runoff data, iii) conducting stepwise regression analysis to investigate scaling behaviours of the observed runoffs in relation to the rainfall characteristics, iv) simulating overland flow and scaling effects by coupling traditional infiltration theory, run-on process and rainfall temporal variations, v) investigating macropore flow contribution to runoff scaling behaviour by measuring vertical pathways of activated macropores with a blue dye experiment at the site, modelling macropore flow in relation to runoff depth, and accounting macropore flow into rainfall-runoff simulations. This is the first study to investigate isolated impact of spatial variability of soil hydraulic conductivity, rainfall parameters, and macropore flow on overland flow scaling behaviour in a burned hillslope. The outcome of this study was partly obtained from field and laboratory measurements and rainfall-runoff monitoring at the field. These measurements and monitoring data were used for rainfall-runoff models parametrisation and verifications. The empirical rainfall-runoff data were collected from multi-scale runoff plots under natural rainfall conditions. The instruments were installed on a severely burned hillslope of eucalyptus forest in southeast Australia. Forty-one rainfall-runoff events were extracted from data collected during the second year following the fire. Strong scaling behaviour was observed for all observed events, seasonally and the whole study period where the rate of runoff declined with increasing plot length. The data from multi-scale runoff plots were used in the stepwise regression models to investigate runoff scaling behaviour in relation to rainfall volumetric and temporal parameters. Stepwise multiple regression analysis showed that generated runoffs and scaling effects were mainly influenced by annual rainfall depth than other rainfall factors while the impact on runoff productions decreases with increasing plot length. Measurements and monitoring data were used for model setup, parameterisation, and verifications of rainfall-runoff models to simulate overland flow scaling effects. The rainfall-runoff simulations provided a very weak demonstration of scaling behaviours with underestimated scaling effects. The simulated scaling behaviours did not improve when spatial variability of soil hydraulic conductivity (CVKs) accounted for the models. This concludes that models with traditional infiltration coupled with run-on process paradigm, and rainfall temporal variability cannot explain the observed scaling behaviour, even where spatially variability of soil hydraulic conductivity (CVKs) is considered. Measurement of activated macropore, water repellency strength and soil water content conducted from the top edge of the hill to downslope on one occasion. No systematic evidence was found regarding reduction in soil water repellency, nor an increase in activated macropore, nor higher water content within distance from uphill. The outcome from these measurements did not support the hypothesis of runoff scaling behaviour to be a result of infiltration increase with distance from uphill. Macropore flow was modelled in relation to runoff depth satisfying pores pressure entry at the point, that is exceeded more frequently with distance downslope due to increased runoff depths. The model consisted of macropore network algorithm, macropore filling when runoff depth exceeded the macropore entry pressure head based on the Young Laplace and Bernoulli equations, and gravity driven vertical flow from fully saturated macropore based on Darcy Law. The macropore flow application was coupled with rainfall-runoff models with traditional infiltration theory, runoff-runon, and rainfall temporal variability. The proportion of simulated macropore flow increased with plot lengths. Simulated scaling effects from models with macropore flow application obtained a better prediction of overland flow scaling behaviour. This concludes that macropore flow is the main factor affecting runoff scaling behaviours in water repellent soil where infiltration mostly occurs through activated macropores and preferential flow. This study supports the theory that macropore flow is a dominant factor controlling overland flow scaling behaviours in burnt dry hillslopes where the soil is strongly hydrophobic. Young-Laplace and Bernoulli equations were found sufficient to calculate macropore filling process in relation to runoff depth, and Darcy Law equation demonstrated continuous flows from full saturated macropore to underneath soil (Hardie et al., 2013, Buttle and House, 1997, Nimmo, 2012, Podgorney and Fairley, 2008). The simulations showed that higher macropore flow is triggered in the longer distance when runoff depth satisfies pressure entry of activated pores. This also explains why the impact of rainfall parameters on runoff productions and scaling decreases with length. The findings of this study are in agreements with earlier findings (Muller et al., 2018, Stoof et al., 2014a, Jarvis et al., 2008, Jarvis et al., 2016, Nimmo, 2012).

Congestion in transport hubs has been a big issue in many population booming cities as the current transport systems are not able to accommodate the increasing surge in travel demand caused by ongoing urbanisation and growth of the world population. The ubiquitous Closed-Circuit Television (CCTV) cameras in transport hubs provide a means for automatic crowd surveillance by utilising accurate and robust image-based crowd analysis models. However, this is an active field of research due to the challenges of occlusion among a large number of individuals and environmental changes, such as lighting fluctuations and changes in context over time. The research presented in this thesis aims to understand and manage the station congestion from the macro and micro perspectives by presenting a framework enabling individual origin-destination estimation and crowd congestion map generation. The main contributions of the study can be outlined as: 1) development of spatial-temporal deep learning models for macro-scale crowd density map generation, and micro-scale pedestrian Origin-Destination (O-D) estimation based on person re-identification in the internet of cameras. 2) a multi-modal framework for crowd analysis under emergency scenarios, especially low visibility situation, using RGBD cameras. 3) linking models and application by developing a web-GIS platform for real-time processing and visualization, thereby demonstrating the applicability of the research in crowd congestion management. This research presents distinct interdisciplinary components with relevance to crowd dynamics as well as image processing and with practical implications for smart station management.

The Arctic is responding to climate change more rapidly and intensely than any other region on Earth. Besides the temperature rise and sea ice retreat, surface gravity waves are becoming more energetic, leading to rapid coastal erosion, affecting wildlife, communities and coastal infrastructures. In response to these changes, this research aims to investigate the metocean parameters across the Arctic Ocean over the past decades by numerical modelling, evaluating their climate and long-term variability. It introduces a systematic assessment of sea ice concentration, winds, and waves by computing their monthly averages, higher percentiles and their long-term trends. The trend analyses demonstrated that the sea ice melting has major responsibility for the general increasing trends in significant wave heights across the Arctic Ocean, as trends in wind speeds were milder compared to trends in waves and sea ice concentration. Finally, non-stationary extreme value analyses were applied in order to assess wind and wave extremes taking into account climate change and seasonal cycles in the estimations. Such an approach becomes essential for the evaluation of extreme environmental variables in the Arctic, where the seasonal sea ice coverage and the ice retreat affect the stationarity of the sea state directly. The results show notable seasonal changes and a consistent increase in extreme waves with the largest increasing rates in the Beaufort and East Siberian seas of approximately 60% in areal-average of 100-year return period for significant wave heights over the past decades. At the same time, extreme winds have only increased by 4% in some regions. Therefore, the changes in extreme winds cannot explain the changes in extreme waves. These results make evident that the sea ice melting has primary responsibility for these dramatic changes in extreme waves.

The effective thermal conductivity of granular geomaterials is the one of the most important parameters in some geotechnical and reservoir engineering applications. Various effective thermal conductivity models have been developed over the years, aiming to predict effective thermal conductivity accurately. However, these models usually predict effective thermal conductivity that are different from the experimental measurements. This may be due to the limited access to the material microstructure and thus limited understanding of its effects on the macroscopic or ‘engineering’ effective thermal conductivity. With the advent of computed tomography and complex network theory, digital samples can be reconstructed based on high-resolution computed tomography images and then microstructure of the samples can be quantified at multiple length scales. At the microscale, three-dimensional sphericity and roundness are selected in this work to describe particle shape based on a critical examination of the literature. They are calculated for each particle in different sands using an in-house developed code. Mesoscale parameters, perhaps with the exception of ‘coordination number’, are still scarcely used in engineering. This issue is addressed by representing the granular materials as either contact networks or thermal networks and applying complex network theory to obtain new mesoscale parameters that characterise the granular assemblies. A network consists of nodes and edges. Specifically, in the contact network, a node represents a particle and an edge means an interparticle contact. Since heat transfers through not only interparticle contacts but also the ‘small’ gaps between adjacent particles, the contact network can be extended to a thermal network by considering the ‘small’ gaps as new edges. The complex network features extracted from these networks can capture the information of particle connectivity or/and contact quality which are essential to heat transfer. Results show that granular geomaterials with higher average sphericity or roundness can render a higher effective thermal conductivity because the two particle shape descriptors have a positive correlation with average coordination number and interparticle contact radius ratio (i.e., the ratio of the equivalent radius of interparticle contact area to the particle radius). Different types of network features can characterise the microstructure from diverse aspects. For examples, the degree is related to contact number, closeness centrality is about the average distance between particles, and betweenness centrality describes the role of a node or edge acting as a ‘bridge’. Many network features can be used as predictors of effective thermal conductivity, especially the features weighted by contact area in the contact network or by thermal conductance in the thermal network. In this work, a single weighted network feature that considers both particle connectivity and contact quality have shown a strong correlation with effective thermal conductivity of different granular materials either generated by discrete element method or digital sands. Some examples of these single features include weighted degree and weighted closeness centrality. The implementation of various advanced tools makes access to microstructure becoming readily and promote a data-driven approach to build effective thermal conductivity models automatically without subjective bias. The multiple characterisations and correlations determined through the thesis allow to rigorously select the input parameter(s) for an artificial neural network model with the capability of predicting the effective thermal conductivity with high accuracy and computational efficiency. The proved feasible data-driven framework from this thesis offers a new paradigm for effective thermal conductivity prediction.

Protection against rockfalls occurring alongside landslides contribute to the major part of the disaster management budget in many counties like Switzerland, Japan and Hongkong. Protective structures are usually built over disaster trajectories to safeguard lives and properties. Reinforced concrete barriers that are fitted with gabions are one common form of installations to provide the protection. Few experimental investigations involving impact testings of a rigid reinforced concrete barrier which was fitted with a gabion cushion cover have been reported in the literature. But these investigations were limited to studying the localised actions of impact. The change of structural response behaviour of the barrier as a whole by the presence of a cushion layer is typically not within the scope of the reported investigations. Design methodologies that have been developed are typically limited to overly simplified calculations based on applying an equivalent static force to the barrier. To fill this knowledge gap full-scale pendulum tests have been conducted by the authors on a barrier that was fitted with a gabion cushion layer. The structural response behaviour of the barrier, contact force and tensile strains in the longitudinal reinforcement were of interests. Results recorded from the tests were compared with results from control experiments which were without the protection of any cushion materials. The introduction of a layer of cushion is shown to be able to have the deflection demand on the structure reduced by more than 70% when the amount of energy delivered by the impact is kept constant. An analytical procedure employing the Hunt and Crossley contact model, Swiss code model and two-degrees-of-freedom (2DOF) system modelling technique is presented for evaluating the flexural response demand behaviour of the cushioned barrier. The proposed analytical procedure is shown to be able to predict the reduced deflection demand with a reasonable degree of conservatism. At the end of the thesis, a simple hand calculation procedure featuring the use of design charts is presented for engineering applications. The procedure is illustrated by a worked example which is based on a realistic rockfall scenario.