This study dealt with shrinkage in multiple correlation coefficients computed for sample data when these coefficients are compared to the multiple correlation coefficients for populations and the effect of the ratio of utilized predictors to original predictors on the shrinkage in R square. The study sought to provide the rationale for selection of the shrinkage formula when the correlations between the predictors and the criterion are known and determine which of the three shrinkage formulas (Browne, Darlington, or Wherry) will yield the R square from sample data that is closest to the R square for the population data.

The use of mineral additions and substitutions in concrete production is highly beneficial to society. When used as substitutes for cement, they provide an alternative destination to polluting waste while contributing to reducing the power consumption and air pollution associated with cement production. This study investigated the influence of different concentrations of additions and curing periods of concrete mixes with water treatment plant sludge ash (CLETA), rice husk ash (CCA) and blast furnace slag (EAF) on total shrinkage and chloride penetration in concrete prepared with high early strength Portland cement. Ten mixes with water/binder ratios of 0.35, 0.50 and 0.65, with wet cure periods of 3 and 7 days were tested. CLETA substitutions for Portland cement ranged from 0% to 30%. Three-component mixes with 20% CLETA and 5% EAF, 20% CLETA and 10% EAF and a four-component mix with 15% CLETA, 5% EAF and 5% CCA were also tested. Shrinkage was measured using a length comparator at 0, 7, 14, 21, 28, 35, 56, 91, 182 and 365 days after removing the sample from the controlled humidity chamber. After 91 days of drying, samples were tested for chloride penetration to check chloride penetration depths in shrinking test samples. Results indicate that even though shrinkage values at 3 days are slightly higher, no significant variation was found for different curing periods. The lowest shrinkage value was found in the reference sample for all tested periods and in the two curing periods. Next, in an increasing shrinkage values, came the following samples: 5%CLETA (5 L), 10% CLETA (10 L), 15% CLETA (15 L), which displayed, for most mixes, a similar behavior to that of 10% CLETA (10 L), followed by mixes 20% CLETA (20 L), 25 %CLETA (20 L) and 30% CLETA (30 L). Chloride penetration resistance increased when the curing time was increased from 3 to 7 days. Three- and four-component mixes showed improved performance when compared to binary samples with CLETA and Portland cement. / O uso de adições minerais e de substituições destas no processo de produção do concreto traz grande benefício à sociedade, por dar um destino a resíduos poluentes e, principalmente, por reduzir o consumo de energia e a poluição do ar gerados pela produção do cimento, ao substituir grande parte desse produto na indústria da construção civil. Neste estudo, investigou-se a influência do teor e do período de cura na utilização de misturas contendo diferentes teores de cinza de lodo de ETA (CLETA), cinza de casca de arroz (CCA) e escória de alto forno (EAF), frente à retração total e à penetração de cloretos de concretos com cimento Portland de alta resistência inicial. Para isso, foram testadas dez misturas aglomerantes, nas relações água/aglomerante 0,35, 0,50 e 0,65, com períodos de cura úmida de 3 e 7 dias. As substituições do cimento Portland por CLETA variaram de 0% a 30%, havendo ainda misturas ternárias, 20% CLETA e 5% EAF, 20% CLETA e 10% EAF, e quaternária, 15% CLETA, 5% EAF e 5% CCA. As leituras de retração foram realizadas com o uso do comparador de expansibilidade nas idades de ensaio, 0, 7, 14, 21, 28, 35, 56, 91, 182 e 365 dias após a retirada da câmara úmida. Após o período de 91 dias de secagem, foi realizado o ensaio de penetração de cloretos por imersão, a fim de analisar a profundidade de penetração de cloretos em corpos de prova em estado de retração. Dos resultados obtidos, constatou-se que, de maneira geral, embora tenham sido observados valores de retração pouco maiores para a cura de 3 dias, não foi constatada variação pronunciada entre os prazos de cura. De todas as misturas investigadas, a que apresentou menor valor de retração foi a de referência, em todas as idades de ensaio e nos dois períodos de cura. Seguido desta, em ordem crescente de retração, vem 5%CLETA (5 L), 10% CLETA (10 L), 15% CLETA (15 L), que tiveram na maioria das misturas um comportamento similar ao de 10% CLETA (10 L), seguido das misturas 20% CLETA (20 L), 25 %CLETA (20 L) e 30% CLETA (30 L). No que diz respeito à penetração de íons cloreto, houve um aumento na resistência quando o tempo de cura passou de 3 para 7 dias. Constatou-se que as misturas ternárias e quaternárias apresentam melhores desempenhos quando comparadas às misturas binárias, compostas por CLETA e cimento Portland.

The selection of variables problem is ussual problem of statistical analysis. Solving this problem via Bayesian statistic become popular in 1990s. We re- view classical methods for bayesian variable selection methods and set a common framework for them. Indicator model selection methods and adaptive shrinkage methods for normal linear model are covered. Main benefit of this work is incorporating Bayesian theory and Markov Chain Monte Carlo theory (MCMC). All derivations needed for MCMC algorithms is provided. Afterward the methods are apllied on simulated and real data. 1

Doctor of Philosophy / Department of Plant Pathology / James C. Nelson / The aim of quantitative trait locus (QTL) mapping is to identify association between DNA marker genotype and trait phenotype in experimental populations. Many QTL mapping methods have been developed to improve QTL detecting power and estimation of QTL location and effect. Recently, shrinkage Bayesian and penalized maximum-likelihood estimation approaches have been shown to give increased power and resolution for estimating QTL main or epistatic effect. Here I describe a new method, shrinkage interval mapping, that combines the advantages of these two methods while avoiding the computing load associated with them. Studies based on simulated and real data show that shrinkage interval mapping provides higher resolution for differentiating closely linked QTLs and higher power for identifying QTLs of small effect than conventional interval-mapping methods, with no greater computing time. A second new method developed in the course of this research toward increasing QTL mapping efficiency is the extension of multi-trait QTL mapping to accommodate incomplete phenotypic data. I describe an EM-based algorithm for exploiting all the phenotypic and genotypic information contained in the data. This method supports conventional hypothesis tests for QTL main effect, pleiotropy, and QTL-by-environment interaction. Simulations confirm improved QTL detection power and precision of QTL location and effect estimation in comparison with casewise deletion or imputation methods.

The current trends in the road transport sector show a growth in axle loads as well as vehicle numbers on all types of roads in highly industrialised countries as well as in developing countries. This increase in axle loads and numbers has forced road agencies to amend their design standards adopting designs that provide roads with higher load bearing capacity. However, the rapid depletion of natural road construction gravel, as well as strict environmental conservation laws have resulted in many agencies in-charge of road construction and maintenance to resort to use of alternative materials that will be economically feasible and environmental friendly. The use of natural pozzolans for stabilising pavement layers fulfils this requirement. This research study was performed with the aim of evaluating the engineering properties of aggregate-lime-natural pozzolan (ALP) mixtures at varying compaction degrees of saturation and to compare them with conventional cement-stabilized aggregates. Two types of pozzolans found in Tanzania were used. The laboratory investigation was carried out in two parts, namely a pilot investigation where the strength behaviour with time, shrinkage and CBR were determined at three degrees of saturation for a washed river sand specimen followed by the main investigation using two different types of sands at four varying degrees of saturation. The study showed that the compaction degree of saturation for ALP mixes plays an important role in their tensile and compressive strengths development regardless of their optimum moisture contents. The ratio between tensile and compressive strengths for ALP mixes was also found to closely obey the relation given by Fulton (2001) for concrete. The ALP mixes were also observed to develop their strength similar to cement mixes with the formation of tobermorite crystals with the additional of water and appropriate activator. Both pozzolan mixes developed significant tensile and compressive strength after 28 days of curing similar to cement mixes. High CBR values for the two ALP mixes were obtained in mixes moulded at degrees of saturation lower than that corresponding to their optimum. Similarly, the shrinkage of the mixes was found to decrease with a decrease in the degrees of saturation. The CBR and shrinkage of the ALP mixes were found to show similar trend to that of the control cement mixes. The ALP mixes showed no significant strength loss with an increase in the fines content in unwashed sand mix in comparison with that of washed sand mixes. No significant strength loss was observed in the ALP mixes as in the control cement mixes at all moulding degrees of saturation. Finally the study concluded that the ALP mixes could be used in stabilization of pavement layers. However, care must be taken in deciding the compaction degrees of saturation as the specifications used in conventional cement stabilization does not necessarily yield desirable strength development in ALP mixes. / Dissertation (MEng (Transportation))--University of Pretoria, 2006. / Civil Engineering / unrestricted

The disposal of fine-grained mining and industrial waste by formation of hydraulic-fill tailings dams is becoming a design and construction activity of increasing scale. In light of the increasing pressure on the mining industry to sustain stringent safety and environmental standards it is becoming more important to gain technical knowledge of the waste problem. The upper layers of the tailings residue dams are in the unsaturated state with the matric suction component contributing to the overall shear strength. The ability to incorporate the matric suction component in shear strength calculations is important to safe design. This research project investigates the use of the mid-plane suction probe to measure matric suction. The results obtained from the probe is used along with various tests to construct a complete soil-water characteristic curve for Mispah gold tailings as well as to investigate suction induced shear strength of drying tailings with depth. The tests were conducted on gold tailings from Vaal Operation’s Mispah tailings dam. The laboratory tests consisted of a trough test, to determine the soil-water characteristics of the gold tailing and also a drying box test that simulated the drying and desiccation of the gold tailings in the daywall. The project concluded that the mid-plane suction probe could be used with acceptable accuracy to determine soil suctions. The model for the prediction of the soil water characteristic curve, derived by Fredlund and Xing (1992), was used successfully to predict the complete soil water characteristics curve for Mispah gold tailings. The equation derived by Vanapalli et al. (1996) was successfully used to calculate both the normal and suction induced shear strength of gold mine tailings using either the volumetric water content from the extracted samples or from the soil water characteristic curve. / Dissertation (MEng (Geotechnical Engineering))--University of Pretoria, 2006. / Civil Engineering / unrestricted

Lors de la construction de structure réalisée en plusieurs phases de bétonnage, les déformations du béton sont restreintes durant son durcissement. Quand le retrait est restreint, des contraintes de traction sont induites dans le matériau et un risque de fissuration est présent. Il est alors nécessaire de modéliser l’évolution des propriétés au jeune âge afin de prédire le comportement de la structure durant le jeune âge du béton. La difficulté réside dans le fait que la modélisation des propriétés du béton doit être basée sur des données expérimentales au jeune âge et que ces données doivent être obtenues automatiquement car le durcissement du béton se produit rapidement pendant les premières heures et les premiers jours. La thèse porte sur l’étude expérimentale et numérique des propriétés au jeune âge des matériaux à base de ciment et plus particulièrement sur le développement des déformations endogènes, le coefficient de dilatation thermique, le module d’élasticité et le fluage propre en compression et en traction. A cet effet, un travail complet a été réalisé à l'ULB et à l’Ifsttar impliquant le développement d'une nouvelle approche avec de nouvelles procédures d'essai et la conception de nouveaux dispositifs d'essai pour générer des données expérimentales depuis la prise du matériau. La méthodologie est basée sur deux méthodes d'essai répété. Pour la caractérisation du comportement viscoélastique d'un béton depuis sa prise, un essai de chargement permanent couplé à un essai avec des chargements répétés de plusieurs minutes est nécessaire. Les déformations endogènes, le coefficient de dilatation thermique et la prise sont caractérisés avec des variations thermiques répétées sur un échantillon de béton. Cette nouvelle approche a été définie sur un béton ordinaire et ensuite étendue sur l’étude de 4 paramètres pertinents : le rapport eau-ciment, l'effet de restriction de l'agrégat sur la pâte de ciment dans le développement des propriétés du béton au jeune âge, la substitution du ciment par des additions minérales et la différence de comportement en traction et en compression. Sur la base de ces résultats expérimentaux, de nouveaux modèles ont été développés pour la caractérisation des propriétés au jeune âge de matériaux cimentaires depuis le temps de prise. Une version adaptée de la modélisation du fluage propre dans le Code modèle 2010 est également proposée / For usual concrete structure built in several phases, concrete deformations are restrained during the hardening process. When shrinkage is restrained, tensile stresses are induced and a cracking risk occurs. Modelling the evolution of an early age set of parameters on concrete is necessary to predict the early age behaviour of concrete structures. The difficulty lies in the fact that the modelling of concrete properties must be based on experimental data at early age and this data must be obtained automatically because the hardening process of the concrete takes place rapidly during the first hours and also the first days. The thesis deals with experimental and numerical study of the early age properties of cement based materials and more specifically the development of the autogenous deformation, the coefficient of thermal expansion, the E-modulus and the basic creep in compression and tension. For this purpose, a comprehensive work was carried out at ULB and Ifsttar involving the development of a new approach with new test procedures and the design of new testing devices to generate experimental data since the setting of the material. The methodology is based on two repeated testing methods. For the characterization of the viscoelastic behaviour of a concrete since setting, a permanent loading coupled to a test with repeated minute-long loadings is needed. Whereas, the autogenous strain, the coefficient of thermal expansion and the setting are characterized with repeated thermal variations on a concrete sample. The new approach was defined on an ordinary concrete and then extended to the study of the following parameters: the water-cement ratio, the restrained effect of aggregate on the cement paste in the development of concrete properties at early age, the substitution of cement by mineral addition and the difference of behaviour in tension and in compression. Based on these experimental results, new models were developed for the characterization of the early age properties of cement based materials since setting time. An adapted version of the Model Code 2010 for the modelling of basic creep is also proposed

South Africa, with its world leadership in the mining sector, and well-developed industrial sector, understandably has many tailings dams of various types. South Africa’s tailings dams are among the largest in the world in terms of delivered tailings tonnages, plan size and height. Obviously tailings disposal from the mining and industrial sectors in South Africa can have a major impact on the environment and the safety of human life if the dam design and tailings deposition process are not properly controlled. In South Africa there is a growing awareness of the importance of the environment and of the safety of the tailings dams. Catastrophes like the Merriespruit Gold Tailings dam failure in February 1994, where 17 people died and widespread devastation and environmental damage was caused, has sparked the renewal of research into tailings dam stability and safety. The rate of rise of tailings dams has an influence on the safety and stability of a tailings dam. If rate at which a tailings dam is built is too high, the dam may become unstable and be at risk of failure. There are many factors that control the rate of rise of tailings dams that are not very well understood. This research deals with sedimentation and desiccation of gold mine tailings. Sedimentation and desiccation are factors that influence the rate of rise. This research looked at how the gold mine tailings behave when sedimentation and desiccation occur. This was achieved through laboratory experiments, which consisted of column settling tests and drying box tests, and field tests. A model that predicts the behaviour of sedimentation and desiccation of tailings was also analysed. It was found that tailings sedimentation occurs very quickly. It was also found that suctions play an important role during the desiccation of the tailings. / Dissertation (MEng (Geotechnical Engineering))--University of Pretoria, 2007. / Civil Engineering / MEng / unrestricted

No description available.

Understanding the fundamental mechanisms of solidification and the relative significance of different parameters governing these mechanisms is of vital importance for controlling the evolution of microstructure during solidification, and consequently, for improving the efficacy of a casting process. Towards achieving this goal, the present work attempts to study the effect of convection and shrinkage on solidification and microstructure formation primarily through the development of computational models which are complemented with experimental investigations and analytical solutions. Convection strongly influences the solutal and thermal distribution adjacent to the solidification interface and affects the growth rate and morphology of dendrites. To investigate this, a numerical model based on the enthalpy method is developed for binary alloy dendrite growth in presence of convection. The model results are validated with corresponding predictions using level-set method and micro-solvability theory. Subsequently, the model is applied for studying the effect of convection on the growth morphology of single dendrites. Results show that the presence of flow significantly affects the thermo-solutal distribution and consequently the growth rate and morphology of dendrites. Parametric studies performed using the model predict that thermal and solutal Peclet number and melt undercooling strongly influence the tip velocity of dendrites. Additionally, an analytical model is developed to quantify the effect of convection on dendrite tip velocity through the definition of an equivalent undercooling. An expression for this equivalent undercooling is derived in terms of the flow Nusselt and Sherwood numbers and the analytical equivalent undercooling values are compared with corresponding predictions obtained using the numerical model. Subsequently, the interaction of multiple dendrites growing in close proximity is studied. It is observed that the presence of neighbouring dendrites strongly influences the thermo-solutal distribution in the domain leading to significant changes in growth pattern. The effect of seed density on the growth morphology is investigated and it is observed that a higher initial seeding density leads to more spherical dendritic structure. Comparison with results from chilled casting of Al-6.5% Cu alloy with and without grain refiners show qualitative similarity in both the cases. The next part of the thesis presents a eutectic solidification model developed using the general enthalpy-based framework for dendritic solidification. New parameters and rules are defined and suitable modifications are made to incorporate the physics of eutectic solidification and account for the additional complexities arising due to the presence of multiple solid phases. The model simulates the presence of buoyancy driven convection and its interaction with the solidification process. i The model predictions are found to be in good agreement with the Jackson-Hunt theory. At first, the model is applied to simulate regular eutectic growth in a purely diffusive environment and it is observed that the model predicts the variation in interface profile with change in lamella width similar to those observed in experimental studies on eutectic solidification. Subsequently, a few case studies are performed to demonstrate the ability of the model in handling complex scenarios of eutectic growth such as width selection, lamella division and presence of solutal buoyancy. It is observed that solutal buoyancy gives rise to flow cells ahead of the eutectic interface facilitating the transfer of solute between the two phases. Apart from forced and natural convection, another important factor affecting solidification is the presence of shrinkage. Currently, solidification shrinkage is mostly modelled using empirical relations and criteria functions. In the present work, a phenomenological model for shrinkage driven convection is developed by incorporating the mechanism of solidification shrinkage in an existing framework of enthalpy based macro-scale solidification model. The effect of shrinkage flow on the free surface deformation is accounted for by using the volume-of-fluid method. The results predicted by the model are found to be in excellent agreement with analytical solutions for one-dimensional solidification with unequal phase densities. A set of controlled experiments are designed and executed for validating the numerical model. The experiments involve in-situ X-ray imaging of casting of pure aluminium in a rectangular cavity. The numerical predictions for solidification rate, free surface movement and temperature profiles are compared with corresponding experimental results obtained from the in-situ X-ray images and thermocouple data. Subsequent case studies, performed using the model, show significant influence of applied heat flux and mould geometry on the formation of shrinkage cavities. The shrinkage flow model provides the foundation for development of a generalized model to accurately predict the formation and morphology of internal porosity. The validated macro-scale shrinkage model is extended to the microscopic scale to study the influence of shrinkage flow on the growth rate of dendrites. Results demonstrate that shrinkage driven convection towards the dendrite strongly influences the solutal and thermal distribution adjacent to the solidification interface and consequently decreases the growth rate of the dendrite. Additionally, an analytical model is developed to quantify the effect of shrinkage driven convection through the definition of an equivalent undercooling for shrinkage flow. The present models provide significant physical insight into various mechanisms governing the process of solidification. Moreover, due to their similar framework, the individual models have the potential to be an effective foundation for the development of a generalized multi-scale solidification model incorporating the presence of important phenomena such as shrinkage and convection.