In the past few decades, several researchers have studied the effects of ASR induced expansion in concrete. Several models have been proposed to model the effects of ASR in concrete. While most of these models focus on plain concrete, there is limited amount of research to model the influence of ASR expansion in reinforced concrete. Additionally, the existing models are complex and difficult to implement for practicing engineers. In this study the shortcomings with the existing models are addressed. A minimalist semi-empirical model is developed to represent the degradation of reinforced concrete due to ASR expansion. The model is validated using historical experimental data. Only two key parameters are needed to represent the expansive behavior, specifically, the maximum unreinforced concrete strain due to ASR expansion and the rise time. Mechanical properties of the reinforced concrete are also needed. From the predicted expansions, it is then shown that it is possible to model the number and spacing of cracks of a partly restrained reinforced concrete beam affected by ASR gels. The model is validated with recent experimental results on large scale reinforced concrete specimens. Predictions agree well with the observed number of cracks.

Partial replacement of cement by metakaolin is known to improve concrete durability, by refining the pore structure to improve fluid transport properties and by reducing the alkalinity of the pore solution to make ASR less problematic. A gap exists in the literature as to the salt scaling performance and magnesium sulphate performance of metakaolin concrete. 10 concrete mixes were cast at w/cm ratios of 0.4 and 0.5. In addition to metakaolin, grade 80 slag was used as well as CSA GU cement. All replacement levels were tested for setting properties, compressive strength, drying shrinkage, sulphate resistance, salt scaling, ASR mitigation and chloride diffusion characteristics according to the appropriate CSA and ASTM standards. It was found that metakaolin showed better performance in magnesium sulphate solutions than slag and that metakaolin concretes attained strength and other durability properties much faster than mixes without it. Keywords: metakaolin, ASR, sulphate, magnesium, scaling, setting, diffusion

No description available.

Les Ranunculales sont un ordre d’angiospermes d’environ 4500 espèces, incluant des plantes communes comme les boutons d’or et les coquelicots. Leurs fleurs sont très diversifiées et ont piqué l’intérêt des botanistes depuis des décennies. Les sept familles de l’ordre sont faciles à reconnaître sur le terrain, mais paradoxalement, certaines familles n’ont pas de synapomorphies et ont des états ancestraux peu clairs, comme les Ranunculaceae (~ 2500 espèces). Cette diversité florale peut être liée à la pollinisation et des innovations ont pu être guidées par cette interaction. De plus, les plantes de ce groupe produisent une grande variété de composés secondaires, certains dont les propriétés sont connues depuis fort longtemps (comme la morphine et le curare). Ces substances pourraient jouer un rôle dans la défense contre les herbivores. Les caractères floraux et chimiques sont donc de bons candidats pour produire des hypothèses adaptatives. Ici nous utilisons des méthodes analytiques modernes pour comprendre les patrons d’évolution expliquant la distribution actuelle de la diversité et l’évolution des fleurs et des toxines de Ranunculales. Un nouveau cadre phylogénétique synthétique a été produit avec des séquences publiées de 144 espèces. Seize caractères floraux et la présence de certains composés métaboliques ont été reconstruits sur cet arbre avec la méthode de parcimonie et la méthode de « Reversible Jump Monte Carlo Markov Chains ». Un nouveau scénario d’évolution florale a été produit, où les Ranunculales avaient une fleur ancestrale trimère avec trois cycles de tépales et où les ancêtres des familles ont évolué en perdant ou en différenciant des cycles du périanthe. Les patrons d’évolution des toxines montrent de la variabilité, certaines apparaissant et disparaissant aléatoirement (comme les saponines), d’autres étant des synapomorphies bien soutenues de certains clades (comme les diterpènes de type aconitine pour la tribu des Delphinieae). Ces résultats offrent une meilleure compréhension de l’histoire naturelle des Ranunculales. En outre, cet ordre est le groupe-frère de toutes les autres eudicotylédones, et a donc une position clef pour comprendre l’évolution précoce de ce clade. Ces résultats vont fournir une compréhension plus profonde des changements floraux et phytochimiques qui ont eu lieu à la base des eudicotylédones, qui contiennent 70% des angiospermes actuelles. / Ranunculales are an order of angiosperms comprising ca. 4,500 species including common plants like buttercups and poppies. Their flowers are highly diversified and have raised the interest of botanists for decades. Each of the seven families of the order is easily recognizable in the field, but paradoxically some families lack floral synapomorphies and have unclear ancestral states, like Ranunculaceae (ca. 2,500 species). This floral diversity may be linked to pollination and innovations may have been driven by this interaction. In addition, the plants of this group produce a variety of secondary compounds, some of which having been known for ages for their properties, like morphine or curare. Those substances could play a role in defense against herbivores. Both floral and chemical characters are thus good candidates to produce adaptive hypotheses. Here we used modern analytical methods to understand the the evolutionary patterns accounting for the extant distribution of diversity and evolution of Ranunculales flowers and toxins. A new synthetic phylogenetic framework of the Ranunculales based on already available sequences of 144 species was produced. Sixteen characters of the flower, and the presence of selected secondary metabolites were reconstructed on this tree using parsimony, and Reversible Jump Monte Carlo Markov Chains. A new scenario for floral evolution was produced, where Ranunculales had a trimerous perianth with three whorls of tepals ancestor, and the ancestors of the families evolved by losing and/or differentiating perianth whorls. The evolutionary patterns of toxins showed variability, some appearing and disappearing randomly (such as saponins), others being well-supported synapomorphies (such as aconitine-like diterpens for the tribe Delphinieae). These results offer a better understanding of the natural history of Ranunculales. Furthermore, this order is the sister-group of all other eudicots, and thus has a key position to understand the early evolution of this clade. These results will provide a deeper understanding of the floral and chemical changes that took place at the base of eudicots, which contains 70% of living angiosperms.

Concrete has proven to be, by far, one of the most reliable materials for the construction of critical infrastructure. However, despite its structural capacity, concrete members are susceptible to damage mechanisms that may decrease its performance and durability throughout its service life. One such mechanism is alkali-silica reaction (ASR), which takes place when unstable siliceous phases present in coarse or fine aggregates react with the alkali hydroxides from the concrete pore solution, generating a secondary product (i.e., ASR gel); this product swells upon moisture uptake from the surrounding environment, leading to cracking and expansion of the affected concrete. In severe cases of ASR-affected infrastructure, structural safety could become a problem, and thus requiring the demolition of affected members. It is, therefore, necessary to adopt effective protocols for the diagnosis and prognosis of aging infrastructure, to ensure its performance over time along with properly planning for rehabilitation strategies, whether required. This work presents a two-stage case study of the S.I.T.E. building at the University of Ottawa for the diagnosis and prognosis of ASR-affected members (i.e., columns) after nearly 20 years in service. The diagnosis phase was conducted with the aim of evaluating the cause and extent of distress and interpreting its impact on the performance of the affected structure. First, a visual inspection was conducted to evaluate potentially damaged members, in order to select the best location for core-drilling. Once ASR was confirmed through petrographic examination, specimens were evaluated through the multi-level assessment (i.e., coupling of microscopic and mechanical assessment). A range of damage was discovered among the examined columns (i.e., 0.03%, 0.05%, and 0.08% expansion). Moreover, evidence of developing freeze and thaw (FT) damage was discovered in columns with greater levels of expansion, raising future concerns regarding the durability and serviceability of members affected by this coupling of damage (i.e., ASR+FT). For the second stage of this project (i.e., prognosis), a novel ASR semi-empirical model was developed with the aim of predicting future ASR-induced expansion and damage in the S.I.T.E. building. The above model was developed and validated (using ASR exposure site data) through the coupling of existing chemo-mechanical macro-models, which were used to predict material behaviour on the structural scale, and novel mathematical relationships for the prediction of anisotropy in the columns. Moreover, the use of the multi-level assessment to predict the mechanical implications of predicted distress was found to enhance the model’s capacity for prognosis and demonstrated important potential for the accurate prediction of multi-level damage in the S.I.T.E. columns.

Learning representations is a central challenge in machine learning. For speech recognition, we are interested in learning robust representations that are stable across different acoustic environments, recording equipment and irrelevant inter– and intra– speaker variabilities. This thesis is concerned with representation learning for acoustic model adaptation to speakers and environments, construction of acoustic models in low-resource settings, and learning representations from multiple acoustic channels. The investigations are primarily focused on the hybrid approach to acoustic modelling based on hidden Markov models and artificial neural networks (ANN). The first contribution concerns acoustic model adaptation. This comprises two new adaptation transforms operating in ANN parameters space. Both operate at the level of activation functions and treat a trained ANN acoustic model as a canonical set of fixed-basis functions, from which one can later derive variants tailored to the specific distribution present in adaptation data. The first technique, termed Learning Hidden Unit Contributions (LHUC), depends on learning distribution-dependent linear combination coefficients for hidden units. This technique is then extended to altering groups of hidden units with parametric and differentiable pooling operators. We found the proposed adaptation techniques pose many desirable properties: they are relatively low-dimensional, do not overfit and can work in both a supervised and an unsupervised manner. For LHUC we also present extensions to speaker adaptive training and environment factorisation. On average, depending on the characteristics of the test set, 5-25% relative word error rate (WERR) reductions are obtained in an unsupervised two-pass adaptation setting. The second contribution concerns building acoustic models in low-resource data scenarios. In particular, we are concerned with insufficient amounts of transcribed acoustic material for estimating acoustic models in the target language – thus assuming resources like lexicons or texts to estimate language models are available. First we proposed an ANN with a structured output layer which models both context–dependent and context–independent speech units, with the context-independent predictions used at runtime to aid the prediction of context-dependent states. We also propose to perform multi-task adaptation with a structured output layer. We obtain consistent WERR reductions up to 6.4% in low-resource speaker-independent acoustic modelling. Adapting those models in a multi-task manner with LHUC decreases WERRs by an additional 13.6%, compared to 12.7% for non multi-task LHUC. We then demonstrate that one can build better acoustic models with unsupervised multi– and cross– lingual initialisation and find that pre-training is a largely language-independent. Up to 14.4% WERR reductions are observed, depending on the amount of the available transcribed acoustic data in the target language. The third contribution concerns building acoustic models from multi-channel acoustic data. For this purpose we investigate various ways of integrating and learning multi-channel representations. In particular, we investigate channel concatenation and the applicability of convolutional layers for this purpose. We propose a multi-channel convolutional layer with cross-channel pooling, which can be seen as a data-driven non-parametric auditory attention mechanism. We find that for unconstrained microphone arrays, our approach is able to match the performance of the comparable models trained on beamform-enhanced signals.

Data-Driven Pronunciation Generation for ASR Maria Obedkova In ASR systems, dictionaries are usually used to describe pronunciations of words in a language. These dictionaries are typically hand-crafted by linguists. One of the most significant drawbacks of dictionaries created this way is that linguistically motivated pronunciations are not necessarily the optimal ones for ASR. The goal of this research was to explore approaches of data-driven pro- nunciation generation for ASR. We investigated several approaches of lexicon generation and implemented the completely new data-driven solution based on the pronunciation clustering. We proposed an approach for feature extraction and researched different unsupervised methods for pronunciation clustering. We evaluated the proposed approach and compared it with the current hand-crafted dictionary. The proposed data-driven approach could beat the established base- lines but underperformed in comparison to the hand-crafted dictionary which could be due to unsatisfactory features extracted from data or insufficient fine tuning. 1

Talking to a computer is hard. Large vocabulary automatic speech recognition (ASR) systems are difficult to use and yet they are used by many people in their daily work. This thesis addresses the question: How is ASR used and made usable and useful in the workplace now? To answer these questions I went into two workplaces where ASR is currently used and one where ASR could be used in the future. This field work was done with designing in mind. ASR dictation systems are currently used in the Australian Public Service (APS) by people who suffer chronic workplace overuse injuries and in the Hansard department of Parliament House (Hansard) by un-injured people. Analysing the experiences of the users in the APS and at Hansard showed that using an ASR system in the workplace follows a broad trajectory that ends in the continued effort to maintain its usefulness. The usefulness of the ASR systems is �performed into existence� by the users with varying degrees of success. For both the APS and Hansard users, they use ASR to allow work to be performed; ASR acts to bridge the gap between otherwise incompatible ways of working. This thesis also asks: How could ASR be used and made usable and useful in workplaces in the future? To answer this question, I observed the work of communicating sentences at the ACT Magistrates Court. Communicating sentences is a process that is distributed in space and time throughout the Court and embodied in a set of documents that have a co-ordinating role. A design for an ASR system that supports the process of communicating sentences while respecting existing work process is described. Moving from field work to design is problematic. This thesis performs the process of moving from field work to design, as described above, and reflects the use of various analytic methods used to distill insights from field work data. The contributions of this thesis are: � The pragmatic use of existing social research methods and their antecedents as a corpus of analyses to inspire new designs; vi � a demonstration of the use of Actor-Network Theory in design both as critique and as part of a design process; � empirical field-work evidence of how large vocabulary ASR is used in the workplace; � a design showing how ASR could be introduced to the rich, complicated, environment of the ACT Magistrates Court; and, � a performance of the process of moving from field work to design.

Fly ash has been used extensively to control deleterious alkali-silica reaction in concrete. The majority of fly ashes can be used to control ASR induced expansion. Fly ashes with high CaO contents are less effective at reducing expansion and fly ashes with high alkali contents can be counter active. Class C fly ashes are less effective at reducing the pH of the pore solution because they are less pozzolanic. The pozzolanic reaction in Class F fly ashes enhances the ability for the hydration products to bind alkalis. This prevents the availability of these alkalis for ASR. This project aims to characterize fly ash in a way that best predicts how it will perform in concrete with an emphasis on ASR. Fly ashes with a variety of chemical compositions were evaluated using a range of analytical and characterization techniques. Research data from several universities were used to correlate their long term data with this project’s accelerated tests. This research aimed at evaluating the mineralogical, chemical, and physical characteristics that most affect the ability of a given fly ash to prevent ASR-induced expansion and cracking. / text

Emirate of Abu Dhabi is located in an arid region, where the main source of fresh water is desalination plants. The vulnerability of desalination plants renders planning for an alternative source of freshwater essential. In this study the feasibility of aquifer storage and recovery in the Liwa area, in Emirate of Abu Dhabi, United Arab Emirates was investigated. Based on operational data collected from the pilot project, the model was set up and calibrated. The calibrated model was used to study the affect of various operational parameters, namely storage duration, pumping rate, screen location, multiple cycle operation and periodic recharge, as well as some aquifer characteristics factors: dispersion and salinity profile. This study can be utilized to optimize the operation of the Liwa ASR project. / text