Civil Engineering - Theses
Permanent URI for this collection
Search Results
1 - 5 of 5
-
ItemHumans’ decision-making during emergency evacuations of crowded environments: behavioural analyses and econometric modelling perspectives( 2017)Modelling evacuation behaviour of humans has become an increasingly important topic due to the growth of urban populations and mass gatherings as well as the increasing frequency of emergency incidents in environments that host large numbers of humans. Emergencies are relatively rare occurrences with high potential impact and safety-related implications. Thus, preparedness for them has the potential to save lives by preventing injuries during the evacuation process and accelerating the overall process and leading people to safety in the least possible amount of time. The research on this topic ultimately aims at developing predictions tools that could facilitate evacuation planning and optimisation by producing measures such as total evacuation times for each given condition. This would enable authorities and planners to evaluate the effectiveness of their evacuation policies, identify potentially problematic locations and vulnerabilities in their venues and propose measures that can accelerate the discharge of occupants should an incidents occurs in the environment. Such applications could range from simply advising occupants as to how they should conduct themselves in case of an incident, to estimating the safe occupancy rate of venues (particularly in mass gatherings and special events) and thus managing the demand accordingly, to optimising the architectural design of the environments in ways that best support the efficient discharge of occupants. The problem is interdisciplinary by nature and has attracted the attention of researchers in various fields, including ergonomics and fire safety, applied physics and mathematics, behavioural sciences and transport engineering. The problem at hand is also highly multifaceted and entails many aspects such as computational capacities and the versatility of the models. The most crucial component of such practice is arguably the accuracy of modelling that is inextricably linked with the humans’ behaviour element and how accurately it can be replicated by the models. Given the safety-related implications of evacuation models, it is of paramount importance to minimise the extent and likelihood of inaccurate estimates that could potentially culminate in misguided designs or suboptimal policies (contradicting the primary purpose for which such models are intended). Ensuring that the behaviour of evacuees in simulated practices are replicated accurately enough is, however, a highly challenging task. The modeller deals with a problem related to humans’ decision-making behaviour which is intrinsically complex, in addition to the fact that the behaviour is this context is particularly rare and is not observed on a day to day basis. This leaves modellers with a paucity of data for model development, calibration and verification purposes. Given that the human behaviour in emergencies is not yet well understood, more often than not, the modellers have resorted to formulating “intuitive” assumptions whose accuracy have yet to be scrutinised based on empirical observations. This has left a range of mysterious theoretical assumptions in this field of research largely unverified and thus subject to debate and scepticism. I argue that the empirical knowledge in this research field has lagged notably behind the theoretical advancements and model formulations, calling for more extensive empirical research in this field in order to bridge this gap. Furthermore, by analysing the existing empirical literature in this field, I argue that the research has been distributed in a relatively imbalanced way in terms of addressing various aspects of human behaviour relevant to this research area. More empirical knowledge has been acquired in relation to the aspects of behaviour that are more convenient in terms of data collection, namely the “walking behaviour” and momentary “collision avoidance” decisions of people. Whereas, higher levels of escape decision making like the directional wayfinding choices or choices of activities are in comparison far less understood. These less explored aspects of behaviour often entail a heightened cognitive load (compared to instantaneous and largely subconscious walking decisions) and pose additional levels of complexity for experimentation, data extraction and modelling. However, I argue in that gaining an accurate understanding of these aspects and developing models that can echo them adequately in the modelling process is at least as important as modelling the walking behaviour from a practical standpoint. In this study, I focused on modelling and understanding the directional (or wayfinding) choices of humans during evacuations that are often referred to as “tactical decisions”. The study is predominantly empirical. In carrying pout this study, a major underlying question was to choose the experimentation method that best suits this problem. As explained in the literature review chapter, I classified the data collection and experimentation techniques in this field to seven major categories, a number of which could be potentially used for the question in hand here. Each method offered certain advantages and disadvantages. I study the problem using two general sources of empirical observations: hypothetical-choice data and data extracted from controlled laboratory experiments with actual crowds (which I refer to as “realistic” choice experiments or often as “field-type laboratory” experiments). The hypothetical-choice data was simply gathered by generating fixed sets of directional choice scenarios, visualising them in the form of simple pictures and surveying a sample of subjects one by one. The choice scenario conceptualise trade-offs between “social factors” and “physical factors” of the environment and elicit the prioritisation of respondents between these factors. This decision trade-off was then imitated in a more realistic simulated evacuation experiment where decision makers had to interact with actual crowds in an actual confining environment for making escape decisions. Their decisions were extracted at the level of individuals using the video-analyses of the experiment footage. The choice data obtained from both contexts were structured in an identical way and were analysed using econometric choice modelling techniques. The findings of this study are two-fold, divided into behavioural findings and econometric-related findings. The analysis results provided novel insight into the escape behaviour of humans, allowing me to revisit a number of conventional theoretical assumptions and statistically test them based on empirical observations. In particular, I examined the assumption of “herding behaviour” and “social attraction and repulsion effects” deeply entrenched in the literature. The conclusion was that these assumptions did not perfectly hold true and could in least terms be regarded as overgeneralisations of complex behavioural phenomena that could affect the accuracy of our predictions. Also the disaggregate nature of our data allowed me to address in depth the question of individual differences in this context that had largely been downplayed by previous studies. The close connection established between the hypothetical and realistic choice data also provided interesting findings about how the responses that people state as to what they would do in hypothetical escape scenarios could actually materialise when they make the actual decisions in more realistic contexts. This connection led to findings that not only did offer insight into the possible relevance of the hypothetical choice methods (or virtual-reality experiments in more general terms) in this particular context (with major implications for choosing research directions for future studies in this field) but also could be of the interest of the researchers in the field of experimental economics and econometrics. The modelling practices reported in this study were performed while considering the prospect of the outcomes leading to practical applications. I intended the models reported in this study to be implementable to evacuation simulation tools. The models, as a result, were kept parsimonious and the parameters were kept fully generic. As a practical application of this study, these models were integrated with a social-force model of walking. This dual-layer model is capable of simulating crowd evacuation process in complex environments that entail the choice of direction. Given that the parameters of the directional choice model convey behavioural interpretations, the model has also the potential to provide behavioural insight into the evacuation behaviour from a system perspective (i.e. based on aggregate measures) through computer simulation and manipulation of the simulated behaviour through varying the level of these parameters. A preliminary analysis of this kind has also been reported in the thesis.
-
ItemSeismic assessment of reinforced concrete walls in Australia( 2017)Some non-ductile reinforced concrete walls in buildings were observed to perform poorly in the 2011 Christchurch earthquake, with most of the lives lost from the event caused by the collapse of buildings that relied on these structural elements for lateral support. Reinforced concrete (RC) walls are widely used throughout the Australian building stock as the primary lateral support elements. It is possible that some of these structural elements would perform poorly in a very rare earthquake due to the low standard of detailing that is currently required in Australia, as well as the low earthquake return period that the Building Code of Australia stipulates for their design. The aim of this research has been to assess the seismic performance of reinforced concrete structural walls, both rectangular and C-shaped, in Australia, a region of low-to-moderate seismicity. The current Australian Standard for Earthquake Actions, AS 1170.4:2007, stipulates earthquake hazard values that are based on a seismic hazard map that is over two decades old. A probabilistic seismic hazard analysis was conducted for most of the capital cities in Australia using the AUS5 model to provide a more accurate prediction of seismic hazard in Australia. The results indicate that for some cities, such as Melbourne, the response spectrum is expected to be higher for large return periods in comparison to the design spectra derived using AS 1170.4:2007. Furthermore, a site response study was conducted using equivalent linear analyses to investigate the amplification of the soil response as classified in AS 1170.4:2007 using a range of ground motions that would be expected in Australia. The primary conclusions from the study showed that there can be a large dependency of the soil amplification on the intensity of the earthquake ground motions for the softer soil classes. Moreover, the low intensity ground motions resulted in a higher spectral shape factor for soil class Be and Ce in comparison to factors derived from the current AS 1170.4:2007. An investigation was undertaken to find the displacement capacity of rectangular lightly reinforced and unconfined walls using a finite element modelling (FEM) program, with emphasis on finding the equivalent plastic hinge length. A Secondary Cracking Model (SCM) was formulated, which is a simple, mathematical model that has the potential to predict if a RC wall has a sufficient longitudinal reinforcement ratio to enable “secondary cracking” to occur. The SCM has been validated by comparison with results from the FEM analyses. Equivalent plastic hinge length equations were derived for the rectangular walls that were observed to form secondary cracking and a single, primary crack, and this can be used to predict the displacement capacity of these walls. This estimate of the displacement capacity assumes that the inelastic rotation that occurs over the inelastic region at the base of the wall can be modelled using an equivalent plastic hinge length over which the curvature is assumed to be a constant value. These estimates of the equivalent plastic hinge length are more appropriate for RC structural walls commonly found in Australia due to the parameters used in deriving them (e.g. mechanical properties of steel, longitudinal reinforcement ratio). Moreover, some expressions for the equivalent plastic hinge length that have derived by previous researchers were found to be inappropriate for the walls analysed in this research; these were particularly inaccurate for walls that do not have sufficient longitudinal reinforcement to force secondary cracks to form. The new expressions provide better estimates of the displacement capacity of lightly reinforced and unconfined walls when compared with recent experimental observations. One of the most widely used and popular cross-sections used in structural design of RC walls is the C-shaped section. There is a paucity of information available on the inelastic behaviour of such elements, and virtually no experimental data exists on non-rectangular concrete walls with inferior details commonly found in regions of low-to-moderate seismicity. An extensive number of nonlinear pushover analyses have been conducted based on FEM to investigate the seismic behaviour of C-shaped walls with detailing commonly found in Australia. Based on the FEM results, the SCM, that has been developed for rectangular walls, was found to be able to predict the potential of a single-crack forming in the walls. The direction of loading and mode of bending was found to be particularly important for the seismic performance of these walls. A non-ductile failure was observed for the majority of the walls investigated due to crushing of the unconfined concrete at the ends of the flanges in the governing direction of loading. Further analyses were conducted in the FEM program but with confined boundary ends to emphasise the importance of such structural detailing in allowing some plastic behaviour to be achieved for the governing direction of loading. The equivalent plastic hinge lengths derived from the extensive number of FEM analyses correlated poorly in comparison to the estimates from a number of expressions that exist in the literature, including a recently developed equation specifically for C-shaped walls. Therefore, equivalent plastic hinge lengths were derived from these results and for each direction of loading. A program has been written in MATLAB to derive vulnerability functions for low-rise, mid-rise and high-rise buildings in Australia that use structural walls as their lateral force-resisting system. The city of Melbourne was used as a template for conducting the analyses, and a dataset of thousands of buildings obtained from the National Exposure Information System (NEXIS) and Census of Land Use and Employment (CLUE) databases was included in the assessment. The displacement capacity of each of the buildings was estimated using a moment-curvature analysis followed by a plastic hinge analysis. A range of artificial earthquakes from GENQKE and real earthquakes from the PEER ground motion database on “weathered bedrock” conditions were obtained. These ground motions were subsequently used in equivalent linear analyses using the program SHAKE2000 to find the site response at the surface of different soil columns from shear wave velocity profiles taken predominantly from sites around Melbourne. The National Regolith Site Classification Map was used to estimate the soil conditions underlying each of the building sites. The acceleration and displacement response spectra resulting from these ground motions were used to represent the seismic demand for different site conditions in the capacity spectrum method and to ultimately estimate the vulnerability of the buildings. Thus, vulnerability functions were derived from the results.
-
ItemBlind bolted moment connections to concrete-filled circular hollow section (CFCHS) columns for high seismic regions( 2017)When used as columns, Concrete-Filled Circular Hollow Sections (CFCHS) offer excellent strength, stiffness, and ductility in addition to an attractive appearance. There is also a construction advantage achieved by the elimination of formwork when compared with the construction of RC columns. In light of the brittle failure of welded beam to column connections experienced in the Northridge and Kobe earthquakes and the inapplicability of ordinary structural bolts when connecting beams to closed column sections, a new blind bolt which can be fixed from the outside of the CHS has been developed. This research employs ONESIDE blind bolts produced by AJAX Fasteners which have been modified to take the advantage of the concrete infill; in particular, single headed anchored blind bolts (HABBs) and double headed anchored blind bolts (DHABBs). However, although there has been considerable previous research, and there is ongoing research on moment connections to square hollow sections using the HABBs and DHABBs, there are no guidelines and clear understanding of the behaviour of moment connections to circular hollow sections using the HABBs and DHABBs. Therefore, this research initially involves both experimental and finite element work on the tensile behaviour of HABBs and DHABBs embedded in CFCHS columns, i.e. single and group behaviour. The possibility of deterioration in the behaviour when cyclic loading is applied is also investigated. Developing a semi-rigid or rigid partial-strength connection suitable for use in moment-resisting frames in regions of high seismicity is the main objective of this part of the research, with a focus on achieving sufficient strength and stiffness of the blind-bolted connection to the column. The strength hierarchy in the connection will be such that the strong and stiff blind bolted connection to the column will be paired with energy dissipating devices between this connection and the beam in a similar manner to the Sliding Hinge Joint developed previously for connections to Universal Column sections in regions of high seismicity. These devices, so-called replaceable buckling restrained fuses (RBRFs), will be automatically activated if an earthquake strikes which is larger than the design based earthquake (DBE) and they will provide sufficient displacement capacity under the maximum considered earthquake (MCE).
-
ItemRoad space optimisation for multiclass and multimodal traffic networks( 2017)Traffic congestion has become a serious concern and hindrance to the prosperity of many societies. Among a variety of solutions two approaches are of significant importance: constructing new roads and bridges to ease traffic congestion and promoting public transport. For the latter, the aim is to provide more space in the heart of cities for public transport (buses, trams, etc) aiming to get more commuters to their destinations. Therefore, two central questions have been addressed in this research; (i) investment in the road construction: given a number of candidate projects associated with construction expenses and a limited budget, what is the best choice of projects. This is known as the road network design problem (NDP), and (ii) transit priority lanes: given a road network, which roads should be selected to provide a lane to be exclusively used by public transport modes such that the overall performance of the transport system is not adversely affected. This problem is called the, “transit priority lane design problem” (TPLDP). For the former, (NDP) a hybridized method consisting of the branch and bound algorithm and Benders decomposition method has been developed. For the latter (TPLDP), the concept of Braess paradox was employed to seek for “mis-utilized” space in congested networks to be utilized by public transport. To this end, a merit index aiming to spot potentially some Braess-tainted roads is introduced first. Then a branch and bound algorithm was developed to find the best subset of the Braess tainted roads that have no adverse impact on the overall performance of the network. This study advances the state of knowledge in the above mentioned problems in five areas: (i) the authenticity of the traffic model is enhanced by subjecting all the analysis to multimodal and multiclass traffic circulation, (ii) the methodologies developed in this study are tailored to real world applications as illustrated with numerical analysis, (iii) a RAM-efficient branch and bound algorithm (BB) has been developed such that the expansion of the BB’s tree structure becomes memoryless, (iv) inclusion of the Braess paradox in the pursuit of the transit priority lane would nullify possible adverse effects on the private modes, and (v) a new method for the capacitated traffic assignment has been developed which is called inflated travel time.
-
ItemTowards improvement of the design of borehole ground heat exchangers( 2017)Ground source heat pump (GSHP) systems utilise ground thermal energy stored at shallow, up to 500 m, depths for heating and cooling applications. The underground elements of these systems, ground heat exchangers (GHEs), provide essential thermal interactions of the systems with the ground. In urban areas, GHEs are typically installed vertically to efficiently use the limited available land. Commonly, such GHEs comprise U-loops of plastic pipes installed in boreholes grouted with cement-based grouts after the installation of the pipes. The installation costs of borehole GHEs are usually the largest component of the capital costs of GSHP systems, so accurate sizing of borehole GHEs is crucial to provide a balance between adequate long-term performance and their financial feasibility. This research aims to contribute to the improvement of the design of borehole GHEs by advancing several aspects of current GHE design practices. Firstly, this work verifies some common borehole GHE models and modelling assumptions against experimental data. To do so, a 120kW commercial GSHP system installed in the Elizabeth Blackburn School of Sciences in Melbourne, Australia, was instrumented to monitor thermal performance of its borehole GHEs and ground-GHE thermal interactions. The system uses twenty-eight 50m deep borehole GHEs to extract and inject ground thermal energy. The GHEs and the adjacent ground were fitted with temperature probes and other instruments. The predictions made by common GHE analytical models are assessed against the 2-year monitoring data collected. This includes predictions of ground temperatures around the GHEs, estimations of borehole thermal resistances and modelling of GHE fluid temperatures. The study suggests that, in general, analytical models of borehole GHEs can be successfully used for simulations of their thermal performance, but the models have to be carefully selected for particular conditions based on their limitations. In addition, this work proposes a method for the estimation of the uncertainty of the GHE design length by considering the uncertainties involved with the selection of design parameters when a particular set of design recommendations is followed. Using the proposed method, a case study is presented where borehole GHEs are sized following a commonly applied design process. The uncertainty of the resulting length of GHEs is estimated and the sensitivity of this uncertainty to the uncertainties of individual design parameters is discussed. Also, possible measures to reduce the length uncertainty, including an in-situ thermal response test, are considered. Furthermore, this work examines the design of borehole GHEs for district hybrid GSHP (district HGSHP) systems. In such systems, two or more buildings share GHEs. This study discusses the benefits of district over individual HGSHP systems and presents a method for the optimisation of borehole GHEs for district HGSHP systems that considers thermal demand regimes of individual buildings. Such optimisation can reduce the total lifetime costs of heating and cooling, capital investments and payback periods of HGSHP systems. The importance of considering the demand regimes in the optimisation is demonstrated through a case study. The case study shows that an optimised district HGSHP system can have significant financial benefits over individual HGSHP systems.