Infrastructure Engineering - Theses

Permanent URI for this collection

Search Results

Now showing 1 - 10 of 494
  • Item
    No Preview Available
    Multi-Modal Traffic Signal Control using Deep Reinforcement Learning
    Yazdani, Mohammad Mobin ( 2023-07)
    The urban population growth has made traffic congestion a huge concern in large cities. To tackle this issue, transport authorities have heavily invested in the expansion of road infrastructure and the development of existing transportation systems. The latter is of great interest since it offers cost-effective and time-efficient benefits over the former. Traffic signal control systems are one of these solutions that play a pivotal role in the management of traffic congestion. Popular systems such as Sydney Coordinated Adaptive Traffic Systems (SCATS) and Split Cycle and Offset Optimisation Technique (SCOOT) have shown their effectiveness in hundreds of cities across the globe. Nevertheless, the logic behind the existing traffic signal optimization methods is mainly heuristic or rule-based. Moreover, the existing traffic sensors (e.g., loop detectors) have been around for more than five decades, limiting the capability of these adaptive traffic signal control systems to improve further. With recent advancements in Artificial Intelligence (AI), there is an unprecedented growth of its applications in diverse fields. Intelligent Transport Systems (ITS) have also enabled the observation of stochastic traffic environments using advanced sensor technologies. AI in ITS has created opportunities for future mobility solutions, especially smart traffic signals. In traffic signal control literature, deep Reinforcement Learning (RL) has been extensively studied which takes advantage of Deep Neural Networks (DNN) to extract complex features. Deep RL interacts with the traffic environment and gains experience to learn the optimal signal timings. Despite showing promising results compared to the conventional methods, the conducted studies have not yet examined the multi-modality of traffic flow and, their challenges when implemented in real intersections. For example, the existing literature has only focused on optimizing vehicles' traffic signals. A multi-modal traffic flow includes vulnerable road users (i.e., pedestrians and cyclists) and public transport (e.g., buses, trams) which needs to be taken into account when adjusting the signals. This thesis contributes to the literature in three areas: 1) we propose a novel deep RL-based traffic signal model to control the vehicles and pedestrian flows in their real setting. For traffic states, the pedestrian volumes are fed to the model as well as vehicle traffic. To fairly distribute the green times, an extended reward function is developed that captures the residual delays due to the real interaction between vehicles and pedestrians. Also, the data at the cross-box area of the intersection is included in the reward function which considers the turning movement delays ignored in signal optimization. The experimental results show the superiority of our model compared to baselines in terms of total user delays. 2) we develop a deep RL-based signal priority method that controls trams and vehicles in a multi-modal traffic environment. Instead of typical Transit Signal Priority (TSP) strategies that heavily prioritize trams over vehicles, we grant a fair priority that benefits both modes. The reward function includes the number of passengers on board the tram and penalizes the tram bunching because of headway deviations. The results indicate a significant improvement in signal priority with minimal impact on the side street traffic, promoting average user speed. 3) We conduct comprehensive experiments to evaluate the performance of deep RL models when using video camera data with limited detection ranges. The methodologies tested in the literature assumed a full availability of traffic data along the road, a scenario that is often unrealistic. We show how effective the deep RL traffic signals can perform compared to actual traffic signals, hence, justifying the implementation of deep RL in the real world. In summary, this thesis develops deep RL-based multi-modal adaptive traffic signal control methods with the consideration of real-world challenges. The experiments are conducted in a simulation environment which are calibrated based on actual signal phasing structures and parameters and, the data collected from the intersection. Then, the proposed methodologies are tested and evaluated over learning-based models and existing gap-based signals with advanced logic. We hope that this thesis is helpful for the next studies in the literature and the implementation of deep RL methods in real intersections.
  • Item
    Thumbnail Image
    Understanding geothermal pavement yields using numerical modelling
    GU, Xiaoying ( 2023-12)
    As the global energy demand escalates due to population growth and the depletion of fossil fuels, renewable energy sources gain increasing attention. Shallow geothermal energy, particularly for space heating and cooling, presents a significant opportunity to contribute to the renewable energy matrix, through energy efficiency (demand) and harnessing thermal energy (supply). Ground source heat pump (GSHP) systems utilise the ground as a heat source (in the winter) or a heat sink (in the summer). GSHP systems take advantage of the moderate ground temperatures to gain efficiency, therefore, reducing the operational cost of space heating and cooling. However, the broader implementation of traditional GSHP systems is often hindered by high capital costs associated with the installation of ground heat exchangers (GHEs) either vertically or horizontally in the soil. To mitigate these cost concerns, this doctoral research investigates energy geo-structures, specifically geothermal pavements, as a novel solution. Energy geo-structures integrate structural elements like piles, slabs, diaphragm walls and tunnel linings with absorber geothermal pipes, serving dual functions and thus reducing initial investments. These pipes form the ground heat exchangers (GHEs). This research emphasises the unique thermal behaviour of GHEs when incorporated into pavements, which are installed at a much shallower depths than conventional horizontal GSHP systems and therefore, exhibit different thermal performance characteristics. This research works utilising advanced numerical techniques to: a) develop a validated finite element model for assessing the long-term thermal performance of geothermal pavements at various scales (field and city scale); b) investigate the potential of integrating sustainable solutions such as permeable pavements and microbial induced calcium carbonate precipitation (MICP) with geothermal pavements; and c) conduct parametric studies on critical factors affecting the efficiency of geothermal pavements, leading to a preliminary design methodology for quick feasibility assessment for different climates. Results obtained from this doctorial work shows geothermal pavements can be an economical and well performed residential space heating and cooling option. Geothermal pavements yield better system thermal performance under a relatively balanced load conditions, such as those encountered in Adelaide, South Australia. Besides, incorporating geothermal pavements with sustainable pavement designs have a promising future. Combining MICP technology with geothermal pavements can improve the overall system performance as well as enhance the structural stability. Overall, this work contributes to the advancement of geothermal pavements, providing a more cost-effective approach to harnessing shallow geothermal energy, thereby, supporting the wide adoption of renewable energy infrastructure.
  • Item
    Thumbnail Image
    Energy Screw Piles: Group thermal interferences and use of Phase Change Materials
    Bandeira Neto, Luis Antonio ( 2023-11)
    The attention and investment towards reducing carbon emissions is growing significantly both in Australia and worldwide. Consequently, the use of renewable energy sources has increased, but several reports indicate that the pace of change needs to rise for the carbon emission reduction targets to be met. Geothermal energy is a recognised source of clean renewable energy, and shallow geothermal direct systems are among the most versatile applications. In these systems, underground heat exchangers are built to transfer heat energy from the underground (i.e., soil and/or rock material) at relatively shallow depths to a Ground Source Heat Pump (GSHP), which upgrades this heat to provide thermal comfort to buildings. The heat transfer within the soil is achieved by circulating a heat exchanger fluid within buried pipes. The structures that hold these pipes are underground heat exchangers. While a GSHP system is considered an economic and reliable thermal comfort system, the capital investment involved in building the underground heat exchangers is often high, hence the investment payout time is typically long. Energy geostructures are an alternative to the traditional underground heat exchangers that can reduce the implementations costs. While in traditional GSHP settings structures are built for the single purpose of holding the heat exchanger pipes, energy geostructures are geotechnical structures that provide structural support while encasing heat exchanger pipes within them. The dimensions and geometry of the geostructure(s) are defined by the geotechnical engineering design; hence, the excavation and concrete costs can be excluded from the GSHP implementation share. Different structures can be utilised as energy geostructures (e.g., piles, retaining walls, tunnel linings), this doctoral thesis will focus on energy piles. Energy piles are arguably the most researched energy geostructure among the presented alternatives. However, industry did not embrace its wide usage yet, even though there are several published works about the topic. The reason behind the popularity of energy piles in comparison to other energy structures is related to its slender shape that resembles vertical borehole heat exchangers (BHEs) traditionally used in GSHP systems. Nevertheless, energy piles are typically shorter in length, requiring a larger number of elements than BHEs to be used and are likely to exchange heat not only with the surrounding soil but also with one another. Moreover, most of the research in energy piles focuses on concrete cast-in-place piles. Aiming to address these gaps, this thesis presents a series of experimental tests undertaken in short energy screw piles, which have smaller diameters than the more researched concrete piles and resemble BHEs, but still shorter. Therefore, one of the tests is undertaken in a group of eight energy screw piles connected in series, which combined have similar length of a BHE (approximately 100 m). The tests are performed in full scale energy structures that are part of a GSHP system under construction in Melbourne, Australia. This field study provides insights of short energy screw piles heat exchanger thermal performance and compares it to a BHE tested in the same site. Complementary numerical analyses assess specificities of the energy screw piles and the thermal interference between the elements. Next, the validated numerical model is used to conduct a study regarding the magnitude of the thermal interference between groups of energy piles connected in series. An extensive parametric analysis is conducted, simulating hundreds of Thermal Response Tests (TRTs) in groups of piles such as the original experiment. The results indicate when (and to what extent) thermal interferences occur when the piles are tested under different number, material and geometry configurations. Moreover, the interpretation of these tests is assessed, first using traditional interpretation methods and next with a new proposed methodology that account for the pile-to-pile thermal interference. Considering the findings on the potential of thermal interference hindering the performance of GSHP systems, this thesis conducts a numerical study on the use of Phase Change Materials (PCMs) as a potential solution to reduce the underground thermal interference between piles. First, the energy screw pile is analysed as a single element, where the original grout filling material is replaced by different sand-PCM mixtures, that besides having different thermal properties also introduce the consideration of latent heat to the system. To solve this new problem setting, the previously validated model is expanded to consider the phase change effect. Finally, the model geometry is modified to consider multiple energy screw piles in one last study with PCM. This analysis involves modelling the energy screw piles filled with grout as the original ones experimentally tested, interleaved with neighbouring piles filled with PCM, so that they act as thermal storage piles. This analysis is conducted using a real thermal load of a whole year, considering an hourly time step, to evaluate both short- and long-term effects. Altogether, each chapter provides a contribution to the energy piles field, with the objective of improving the knowledge of this emerging technology, which is expected to increase the confidence of geothermal designers and therefore lead to a wider utilization. Ultimately, extensive efficient application of energy piles will complement other renewable energy technologies and help bringing society to a more sustainable future.
  • Item
    Thumbnail Image
    Irrigation Benchmarking to Boost Water Efficiency and Productivity Using Remote Sensing
    Gao, Zitian ( 2023-07)
    Irrigation is an important component of freshwater consumption. Inefficient water use has been a long-standing problem in many irrigation districts, leading to large amounts of water waste and economic loss. Improving irrigation efficiency can help increase irrigation sustainability and improve food security under climate change. Irrigation benchmarking is a practical management tools that can support the evaluation of irrigation performance and provide valuable insights to inform more efficient irrigation management. However, the application of irrigation benchmarking is often hindered by limited data availability, especially at the field or farm level over a large region. At present, most irrigation benchmarking relies on data obtained through traditional methods such as surveys, which is time-consuming and thus challenging to be consistent and continuous for a long time. Recently, the rapid development of remote sensing (RS) offers new opportunities to obtain spatially explicit data over a large spatial extent with significantly lower human resources. RS data can be a useful indicator of crop distributions, photosynthetic activity, and health and biomass accumulation; potentially filling in critical gaps in inputs to irrigation benchmarking. Furthermore, most existing benchmarking studies have focused on assessing and comparing irrigation performance, with few exploring the underlying drivers of variation in irrigation performance. While benchmarking (comparison) shows how well a management unit performs, the identification of key drivers provides a deeper understanding of reasons for specific levels of performance, which could help facilitate the adoption of new strategies to improve efficiency of poorly performing units. This PhD research aims to develop an RS-based irrigation benchmarking system that can assist in improving irrigation water use at the farm and irrigation district levels. This research is conducted in the Coleambally Irrigation Area, a critical irrigation district in south-eastern Australia. The benchmarking system developed includes 1) comprehensive RS-derived information of irrigation activity, crop types, and crop production and combines it with irrigation water use information to undertake irrigation performance assessment and comparison across farms; 2) development and use of data-driven models to improve understanding of the spatial and temporal patterns of irrigation performance and to identify key drivers of the variation, and 3) informative suggestions based on benchmarking results to improve irrigation management. The key novelty of the benchmarking system developed is its capacity to retrieve cost-effective, high-quality and spatially-explicit information, along with a data-driven approach to interpret this information to make practical recommendations on improving irrigation practices. The benchmarking system developed reduces reliance on ground data and can be fully automated, meaning that it is transferable to other irrigation districts with water use data. Chapters 3-7 explain how the individual components of the benchmarking system achieves these contributions. First, field-level irrigated field and crop type mapping were constructed over the entire irrigation district (Chapters 3 and 4), providing essential geographic information for the benchmarking system. Chapter 3 developed a pixel filtering method and two dynamic irrigated field classification models, allowing irrigated fields to be accurately classified over multiple years using Landsat-based Normalized Difference Vegetation Index (NDVI) time series (Kappa coefficients > 0.9). Chapter 4 further developed the irrigated field maps to include crop types within the cropping season using crop classification models with Landsat spectral band reflectances as input. Three types of summer crops (corn/maize, cotton and rice) were classified with good accuracy (overall accuracy > 95%). In addition, the increase in model accuracy as the season progressed was also evaluated in Chapter 4, providing insights into the earliest time that it is possible to obtain accurate crop maps. Four quantitative experiments were designed to optimise the within-season crop classification accuracy by adopting appropriate training sample strategies. Although not directly linked to irrigation benchmarking, this within-season crop mapping could be used in future studies for mid to late-season crop and water management. Chapter 5 calculated and compared the farm-level relative irrigation supply (RIS) in the Coleambally Irrigation Area over multiple years. RIS is the ratio of crop irrigation supply to crop net irrigation demand (i.e., crop evapotranspiration minus effective rainfall). RS was used to estimate three key inputs of the RIS, namely cropping area, crop coefficients (Kc) and the start and end dates of a season (SOS and EOS). Results showed that over-irrigation (RIS > 1) existed on many farms. In addition, most farms tended to have consistent RIS values when growing the same type of crops over the years. The uncertainty in the estimated RIS was also evaluated, with uncertainty in the cropping area being the primary contributor. More importantly, the cropping area estimated based on RS was found to be more accurate than farmer-reported cropping areas. This indicates the feasibility of using satellite data to obtain reliable information for irrigation benchmarking. In Chapter 6, both field- and farm-scale crop yields for cotton are estimated using remote-sensing imagery and combined with irrigation delivery records for benchmarking irrigation water use efficiency (IWUE) and gross production water use index (GPWUI). A machine learning model was developed based on ground truth yields of cotton and Landsat imagery. Results showed that cotton yields could be predicted with reasonable accuracy at the field level (R2 = 0.63). However, prediction accuracy at the field scale might decrease if a model was developed on sub-field-scale samples (such as from harvester yield maps) or applied to an unseen year. The second part of this chapter combines the predicted yields with irrigation water usage for farm-level irrigation efficiency benchmarking. The results indicated that IWUE for cotton ranged from 0.88-1.55 bales/ML and GPWUI varied between 0.76 and 1.10 bales/ML during the studied period. These results show very good water use efficiency on Australian cotton farms compared with the global average (GPWUI = 0.48 bales/ML). Chapter 7 identified key drivers of variation in seasonal irrigation water usage at a crop-farm-year level. It developed a Bayesian hierarchical modelling framework that allows 1) key drivers of seasonal irrigation water usage to be identified for individual crop types and 2) seasonal irrigation water usage for all crop-farm-year samples to be estimated. Results showed that seasonal irrigation water usage was primarily influenced by irrigation practices and soils, and different subsets of irrigation and soil factors can be used to predict it with reasonable accuracy (R2 = 0.62). This thesis demonstrates that RS can provide reliable data on irrigated cropping area, crop type, crop water requirement and crop yield for irrigation benchmarking with limited human effort and costs. Some variables, such as the irrigated cropping area, can be estimated more accurately from RS than surveys. Farms tend to have consistent RIS values when growing the same type of crops across years, and irrigation in excess of theoretical plant water requirement (RIS>1) is common. This emphasises the need to re-evaluate irrigation usage in poorly performing farms to reduce water waste in the long term. For example, farms with unsatisfactory RIS (e.g., RIS>2, which means the supply is double the plant water requirement) could re-evaluate their irrigation practices and methods. Interannual variability of IWUE and GPWUI for cotton farms is observed, but generally, IWUE and GPWUI in the study area is higher than the global average, indicating a high cotton water use efficiency in Australia.
  • Item
    Thumbnail Image
    Rapid assessment of asymmetrical Reinforced Concrete buildings in regions of low to moderate seismicity
    Xing, Bin ( 2023-10)
    Reinforced concrete (RC) buildings make up the majority of building stock in capital cities in Australia. Design consideration for seismic loading was not mandated in most states of Australia until 1995. Prior to 1995, buildings were designed for gravitational and wind loads only. The structural elements of the buildings were commonly designed without the consideration of ductile detailing. However, despite being located in a stable continental region, Australia, on average, experiences two earthquakes with a magnitude larger than 5 each year and one earthquake with a magnitude of 6 every five years. The casualties, damage and economic loss from the Newcastle earthquake in 1989 have highlighted the potential earthquake risk to Australian society. There is clearly a need to develop a methodology to assess the potential vulnerability of existing RC buildings. This research aims to develop a rapid seismic vulnerability assessment methodology. A methodology has been developed based on a three-level approach based on visual screening and simple analyses. An extensive review of the literature and Standards code of practice was conducted to identify vulnerable features in RC buildings and the associated limits. The outcomes of the literature review have been used as a basis for the visual screening approach developed as a part of the methodology. Also central to the seismic vulnerability assessment is the ability to establish the torsional stability of RC buildings and provide estimations of the displacement demand of the buildings, including buildings featuring horizontal irregularities. Parametric studies based on single-storey and multi-storey building models were conducted to investigate parameters that affect the torsional stability and displacement demands of asymmetric RC buildings. Results from the parametric studies were integrated to develop a simple analysis procedure to provide estimations of displacement demand of RC buildings with asymmetry. The developed vulnerability assessment methodology has been illustrated by a case study of multi-storey buildings. The outcomes have demonstrated the robustness of the developed methodology and its potential use, especially when a large number of buildings are to be assessed.
  • Item
    Thumbnail Image
    A Spatial Causal Inference Analysis Framework
    Akbari, Kamal ( 2023-10)
    The growing interest in causal inference in recent years has led to new methodologies and their applications across disciplines and research domains. Yet, studies on spatial causal inference are still rare. Causal inference on spatial processes faces additional challenges, such as spatial dependency and spatial heterogeneity. These challenges can lead to spurious results and, subsequently, incorrect interpretations of the outcomes of causal analyses. This thesis investigates the importance, common challenges, applied methods, and gaps in spatial causal inference and aims to provide a spatial causal inference framework based on the characteristics of spatial processes. In the first step, this study systematically reviews the relevant literature for spatial causal inference. This review presents a snapshot of the state of the art in spatial causal inference and identifies the methodological gaps, weaknesses, and challenges of current spatial causal inference studies. This study systematically categorises and presents potential causal structures based on combinations of four types of spatial interdependence in outcome variables, treatment, explanatory variables, and error terms in the data generation processes of spatial data. This study proposes a framework to obtain a balanced matching of treated and control groups in a causal analysis of a geographical process. Designing matching methods for spatial data reflecting static spatial or dynamic spatiotemporal processes is complex because of the effects of spatial dependence and spatial heterogeneity. Both may be compounded with temporal lag in the dependency effects on the study units. Current matching techniques based on similarity indexes and pairing strategies need to be extended with optimal spatial matching procedures. The results of this thesis show that the major gap in spatial causal inference is the need for a systematic framework. To account for this problem, this thesis proposes a systematic framework for spatial causal inference. This framework provides a decision tree based on the main components of spatial causal inference and can help researchers working on spatial and geographical problems better understand the potential procedures and solutions for their studies. The framework can identify the building blocks of the tools necessary to perform causal analyses and how these tools should be used in the studies of geographical processes. To demonstrate the application of the proposed framework, two real-world case studies, the wayfinding process and house prices, are evaluated. The causal link between human mobility and spatial information needs is assessed. Here, a methodological framework for causal inference on observational individual-level spatiotemporal data is proposed. This study demonstrates how this framework enables a deeper understanding of causal mechanisms linking wayfinding information needs and environmental familiarity. The results show the causal effects of information needs on navigation systems' user information behaviour and the sensitivity of these results to operationalising causal assumptions. Several potential causal structures are evaluated to estimate the effect of opening three new train stations on house prices in the northern metropolitan region of Melbourne, Australia. These models are applied on different scales and settings to show the application of the proposed framework for spatial causal inference and the sensitivity of the results to the study design, the aggregation level of units, and the type of causal structure. This study uses specific use cases to demonstrate the practical application and effectiveness of the proposed framework. The results of the causal analysis for the case study can support geographic policy analysis, analysis of the effects of infrastructure projects, and the design of spatial recommendation systems for users. In general, this thesis contributes to the domain of spatial causal inference. The results of this thesis can be used in answering causal questions in spatial processes and analysing and estimating causal effects more accurately. The proposed framework for spatial causal inference can be applied and extended by researchers working with spatial data from various disciplines. The results show the importance of causal structures and applied settings on the results of spatial causal analyses and can also provide helpful insights about causal inference on spatial processes.
  • Item
    Thumbnail Image
    Impact of Internal Erosion on Soil Fabric and Mechanical Behavior
    Dong, Chen ( 2023-08)
    Internal erosion is a common phenomenon which involves in the formation of voids within soil matrix induced by the removal of particles by seepage flow. It is believed as the second most common cause of levees and earth dam failures. According to literatures, internal erosion accounts for around 48% of embankment dam failure. Internal erosion is classified into four types based on erosion initiates, progress, and location: concentrated leak, contact erosion, suffusion and back erosion. Among the four erosion types, suffusion, in which case finer particles are eroded out of the voids between coarser particles by seepage, is focused in this research. The potential hazard of suffusion is that it might change the soil structure and thereby impact the mechanical performance of soil matrix. Internal erosion has drawn attention from the field of geotechnical for decades from the view of field tests, laboratory tests and recently numerical modelling. Several attempts have been conducted to find out the impact of suffusion on the stability of soil structure, soil stiffness and strength by experimental approaches. However, due to the disunity and inconsistency of testing material, dimension of testing apparatus and lab testing methods, only a few convictive results are derived, hence it is still less understood in the field of internal erosion especially related impact on mechanical behaviour of soil. Considering the large scale of dam failure, it is difficult to carry out field tests; and since incomputable finer particles participate in the particle migration during the process of internal erosion, numerical modelling cannot accurately simulate the real case of erosion because of uncertainty of erosion time, soil characteristics. This research would carry out a series of laboratory tests with an erosion testing system to find out the relationship in between of fine particle erosion and mechanical behaviour of cohesionless soil. Small strain shear modulus (G0) is selected as the representative of soil mechanical performance. G0, as one essential parameter to evaluate soil structure interaction, significantly contributes to earthquake engineering design. It draws a lot of concerns but still awaits to be clearly understood as a stiffness parameter of sand. Initially, literature review is conducted focusing on previous laboratory erosion tests and laboratory geotechnical tests coupling with bender elements to gain knowledge and find out research gaps. Secondly, a rigid wall permeameter is remoulded to embed bender elements for measurement of G0. Then a set of erosion system is set up to carry out erosion tests providing hydraulic force and vertical loading. Thirdly, before erosion test, gap-graded soil mixture sample is prepared in cohesionless soil with appropriate method to achieve full saturation. Fourthly, a series of erosion tests are carried out with varying potentially impactive parameters while measuring G0 with a signal analysis system in the meantime. This research enhances the understood of the impact on soil fabric and small strain shear modulus due to internal erosion. It contributes to better understanding of the stability of hydraulic buildings and provide valuable evidence for future internal erosion studies.
  • Item
    Thumbnail Image
    3D Data Modelling for Underground Land Administration
    Saeidian, Bahram ( 2023-09)
    Rapid urbanisation and limited land availability have led to increased use of underground spaces over the last decades. Underground assets such as utilities, tunnels, and walkways are essential for the efficient planning and functioning of cities. This increased utilisation of subterranean space has created a rise in its value, highlighting the significance of legal and physical dimensions of underground environments. Underground land administration (ULA) is critical for sustainable urban development. Digital data plays an underpinning role in managing underground spaces. From ULA's perspective, there exist three important underground data elements including physical, legal, and survey data. Physical data includes the spatial and semantic information about the physical reality of underground assets while legal data defines the spatial and semantic information about the ownership spaces and boundaries associated with these assets. Finally, survey data elements are required to define underground legal boundaries and connect underground legal spaces to a geodetic survey network. Underground land information currently exists in silo and fragmented 2D environments and requires extensive consolidation before it becomes usable. In the current practices, 2D survey plans and maps are used to manage and communicate underground physical, legal, and survey datasets. These 2D approaches do not integrate the physical location and legal ownership of underground assets in a common data environment, leading to critical challenges related to data communication and management when planning and developing underground projects. To minimise the impacts of underground projects on third-party properties and assets, several pages of 2D floor plans, cross-sectional diagrams, isometric representations, and textual notations should be reviewed to obtain a consolidated representation of physical, legal, and survey data, which is a cognitively challenging task. In addition, legal data are stored, managed, and communicated at a parcel scale, and there is no 3D land administration data model at a city scale for underground areas. 3D data modelling is one of the first and fundamental steps towards implementing a 3D integrated digital model of underground assets and the rights, restrictions, and responsibilities (RRRs) associated with these assets. Therefore, this PhD research has developed a new 3D data model at a city scale to support integrated management of underground physical, legal, and survey data elements. This data model is based on the CityGML standard as a prominent 3D city-scale data model that is used to represent objects with respect to their 3D geometries and semantics. An application domain extension (ADE) for CityGML 3.0, which is called VicULA ADE, was developed at the conceptual, logical, and physical encoding levels to store, manage, and communicate 3D underground land information at a city scale. The VicULA ADE includes three modules to model geometries, semantics, attributes, and relationships between various types of underground physical assets, legal spaces and boundaries, and survey elements in an integrated 3D environment. The VicULA ADE was implemented and prototyped for several complex underground case study areas to demonstrate the feasibility and practical integration of physical, legal, and survey data elements. The findings show that the VicULA ADE provides a new solution for the integrated management of underground datasets. This integrated 3D data model improves the communication, understanding, and interpretation of underground physical, legal, and survey data and provides a more understandable and superior representation of underground land information compared to current fragmented and isolated 2D approaches. The VicULA ADE also plays an underpinning role in use cases of ULA beyond land registration, such as planning and development of underground projects and underground asset management. Therefore, integrated representation of underground land information using the VicULA ADE can potentially offer significant benefits such as reducing legal disputes, economic losses, and social issues in underground space management. This research proposes a new technical approach for consolidating physical, legal, and survey data associated with underground assets in a 3D city model. This approach addresses the limitations of previous studies which were limited to only physical, legal, or survey data at a building or parcel scale. Additionally, the CityGML VicULA ADE provides a wide range of semantic relationships between underground physical, legal, and survey data elements. The new implementation pipeline developed in this research comprises the key steps for transforming underground physical, legal, and survey datasets to an integrated 3D underground model, which provides valuable knowledge inputs for future studies in 3D underground data integration. In particular, the 3D integrated underground data model is developed based on the specific requirements defined in the Victorian jurisdiction of Australia. However, the proposed approach can be replicated in other jurisdictions by adjusting the data requirements for managing underground land information.
  • Item
    Thumbnail Image
    Behavior And Design of Innovative Coupling Composite Shear Walls and Framing Systems
    Tran, Quang Hau ( 2023-09)
    Composite frames that comprise of concrete-filled steel tubular columns (CFST) and concrete-filled steel plates shear walls have been increasingly utilized in tall buildings due to their superior merits. However, these composite structures are still mainly designed by the member – based design approach in which the structure is considered to fail if one element fails leading to a conservative design. Therefore, this study focuses on developing a framework for the system-based design of this kind of structure using nonlinear simulation and reliability analysis. Firstly, the numerical model is developed in OpenSees to investigate the behavior of high-rise composite buildings with composite shear walls and concrete-filled steel tubular (CFST) columns. In the model, the geometric and material nonlinearities of structural elements are captured by utilizing fiber beam – column elements where the rigorously modified material stress-strain relationships are assigned. Besides, the confining effect of the concrete core, the semi-rigid connections and the coupling effect of composite shear walls are also taken into consideration. Based on the developed model, the case study of a 42-storey composite building is conducted to provide a thorough understanding about the behavior of this kind of building. It shows that this innovative building has higher loading capacity and significantly reduces the dimension of structural members (up to 50%) when being compared with the conventional RC building at the same loading capacity. Through the validation with the test data, the suggested constitutive laws have shown their simplicity and high accuracy since the value of the model error is only around 7%. In addition, the simulation results also indicate that the model can capture well the nonlinear behavior of tested specimens including the failure and the formation of plastic hinges of coupling composite shear walls implicitly. Secondly, an effective reliability analysis procedure is developed to propose the system resistance factors for the system design of steel-concrete composite frames that comprise of concrete-filled steel tubular (CFST) columns and composite beams. Advanced analysis is employed to predict the ultimate resistance of frames using fibre beam-column elements in OpenSees. The obtained predictions of the load-carrying capacity of frames compare well with experimental results with the mean value of the test-to-prediction ratio around 1.027 and the coefficient of variation (CoV) of 8.4%. Both Monte Carlo (MC) and subset simulations are used in the reliability analysis. The uncertainties of model error, geometric and material properties, and external loads are included to predict the system reliability index. Different frame configurations are considered. The results of the reliability analysis show that the system resistance factors of the investigated frames for both US and AS codes are quite similar. In the case of gravity load, the system resistance factor is from 0.78 to 0.90, whilst this value for the case of combined wind and gravity load is from 0.8 to 0.95. Finally, a pioneer framework for the system design of composite frames with composite shear walls is proposed. A C++ source code for a new material is developed in OpenSees to capture the behaviour of semi-rigid connections. This material is implemented in the proposed numerical model to conduct nonlinear simulation of CFST structures. Then, a Matlab code based on subset simulation is programmed for the reliability analysis. This code rigorously considers uncertainties of variables such as model error, geometric and material properties or external loads of both US and AS standards. Finally, case studies of frames with composite shear walls are conducted by using the developed framework to propose system resistance factors of the investigated frames. These proposed factors can be a great source of reference for the system design of CFST tall buildings with composite shear walls.
  • Item
    Thumbnail Image
    Investigate the Inflammatory Response in Early Stage of Fracture Healing Under Various Disease Conditions
    Zhang, Enhao ( 2023-10)
    Secondary fracture healing is the predominant method of bone repair. The inflammatory phase, which is the initial stage, is vital to this process. Following a fracture, immune cells quickly migrate to the site, serving two purposes: clearing cellular and tissue debris resulting from the injury and releasing a cascade of cytokines to initiate the healing process. While an inflammatory response is essential for effective bone repair, uncontrolled inflammation can impair the healing outcome. Chronic inflammatory conditions, like diabetes, may intensify and prolong the inflammatory reaction at the fracture location. An overabundance of immune cells and cytokines has been found to disrupt the growth and differentiation of mesenchymal stem cells, resulting in a healing failure. The mechanical stability of a fracture site can be influenced by various factors, including changes in bone material properties due to osteoporosis, the choice of internal fixation methods, and the physiological stress exerted on the fracture. An unstable fracture site can lead to recurring microtraumas. The constant breaking of newly formed blood vessels at the site attracts more immune cells and releases additional inflammatory cytokines, amplifying inflammation. At present, the correlation between factors contributing to fracture instability and inflammatory responses remains unclear. The impact of different inflammatory conditions,whether induced by mechanical instability or chronic inflammatory diseases,on fracture healing outcomes is yet to be comprehensively understood. At first, this study developed a computational model designed to simulate the TNF-alpha mediated inflammatory response during early fracture healing. We validated this model using data from previously published experimental studies. Once validated, we employed the model to examine the effects of exacerbated inflammation, as seen in diabetes, and the absence of TNF on the proliferation of MSCs and subsequent fracture healing stages. Subsequently, we investigated the interplay between mechanical instability and the inflammatory response during the fracture healing process. To link interfragmentary strain with the inflammatory response, we developed a computational model that illustrates the impact of platelet-derived growth factor (PDGF) released from ruptured blood vessels on macrophage migration and subsequent fracture healing. This model was validated with experimental data from earlier research. The validated model was later used to investigate the bone healing process under mechanically stable or unstable condition with different inflammatory conditions. In further exploration, the study investigated the consequences of risk factors potentially impacting the stability of the fracture callus on both the inflammatory response and the proliferation and differentiation of bone cells. We introduced a numerical model that integrates a 2D fracture callus model with a 3D fracture fixation model. This was constructed to study how osteoporosis-induced alterations in bone material properties affect the interfragmentary strain in tibia fractures stabilized with a locking compression plate (LCP). This LCP fracture fixation model was then adapted to analyze the consequences of varying fixation strategies and loading rates on the fracture site's stability. We validated the model using experimental data sourced from mechanical tests conducted within this study. Lastly, the validated model was employed to forecast the uptake of inflammatory cells and cytokines into the fracture callus, as well as predict fracture healing outcomes. This research has shed light on the intricacies of the inflammation-mediated fracture healing process. It has unveiled the impact of several factors on the inflammatory response during bone repair. Key discoveries include: 1. There appears to be a specific optimal concentration of TNF-alpha in the fracture callus that promotes early-stage healing. Any significant deviation from this concentration, whether due to an overproduction or a deficiency of TNF-alpha, can potentially hinder the healing outcomes. 2. When diabetes and mechanical instability coexist, they can substantially disturb the standard processes of early-stage inflammation. This disturbance can transition acute inflammation into a chronic state, characterized by a persistently heightened TNF-alpha pathway. 3. Osteoporotic fracture calluses can impede the healing process by obstructing the growth and differentiation of mesenchymal stem cells. Moreover, when osteoporosis and diabetes co-occur, they can profoundly jeopardize fracture healing outcomes. 4. During the initial week of fracture healing, a lower frequency, such as 1 Hz, amplifies TNF-alpha production. Regarding osteoporotic fractures, the TNF-alpha concentration in the fracture callus markedly decreases under high loading rates. Especially, fractures with a more flexible configuration demonstrated increased TNF-alpha output. In contrast, the growth and differentiation of MSCs were promoted in fractures subjected to higher loading rates. Importantly, an increased bone-plate distance (BPD) and an extended working length (WL) were found to markedly amplify the inflammatory response, potentially acting as adverse factors in the bone healing process, particularly for patients with diabetes and osteoporosis.