The estimation of a particular matrix of coefficients of a second-order polynomial model was shown to be important in Response Surface Methodology (RSM). This led naturally to designing RSM experiments for best estimation of these coefficients as a primary goal. A design criterion, D_S-optimality, was applied to several classes of RSM designs to find optimal choices of design parameters. Further, previous results on D-optimal RSM designs were extended. The designs resulting from the use of the two criteria were compared. Two other design criteria were also studied. These were IV, the prediction variance of ŷ integrated over a region R, and IV*, sum of the variances of ∂ŷ/∂<u>α</u> again integrated over R. Three different choices of the region R were used. The object of the study was not only to identify optimal choices of design parameters, but also to compare the resulting designs with those obtained using the determinantal criteria. An extension of a method for constructing D-optimal designs was used to construct D_S-optimal central composite designs. This involved viewing the design points as having continuous weights. D_S-best central composite designs were constructed either analytically or numerically for a fixed axial point distance. The results of previous work by other authors were extended for D-optimality by varying the axial point distance. Other design classes studied were Box-Behnken, equiradial, and some small composite designs. The novel study of IV and the extended IV, called IV*, was done for each of the four design classes mentioned previously. The results of the study were presented graphically, or tabularly. The best designs according to IV and IV* were compared with the D_S-best designs. Composite designs performed well in all criteria, with the central composite designs performing best. The Box-Behnken and equiradial seemed to suffer from a lack of flexibility. The D_S-best designs agreed well with the designs suggested by the IV* criteria. / Ph. D. / incomplete_metadata

Narrow gaps in overlapping structural steel surfaces are problematic when being hot-dip galvanized due to the potential for trapped cleaning solutions between the surfaces. A seal-weld is often used to prevent the cleaning solutions from penetrating this gap. However, these welds are not necessary used for strength, and add fabrication costs because of the additional weld. The purpose of this research is to provide alternatives, which fall under two major categories, to the seal-weld fabrication process. The first one was motivated by the steel fabrication industry and uses a commercial silicone caulk to seal the narrow gap instead of a seal-weld. The second was motivated by the galvanizing industry and increases the narrow gap to a minimum of 3/32 in. to allow free flowing of the liquids including viscous molten zinc. 45 specimens in six different overlapping configurations were tested. Three experimental tasks were performed as part of this research: two different types of silicone caulks were used to partially substitute the seal-weld to prevent fluid penetration; an accelerated corrosion test was performed to determine the long-term corrosion resistance of each configuration; and a coating layer evaluation was performed to investigate the bond of the metallurgical layer between the steel and the coating. Results indicate that silicone only partially prevented the penetration of the cleaning solutions into the gap but performed poorly when fully galvanized. Also, the accelerated corrosion and coating investigations indicated that the suggested caulks and the 3/32 in. gap were not as efficient as the seal-weld solution. / Master of Science / Hot-dip galvanizing is a corrosion protective process for steel. The process involves dipping the base material into a series of different baths. The first set of baths are the cleaning baths and the last bath is the zinc coating bath. In the cleaning baths, certain chemicals are applied to clean the steel surface from oxides and rusts and to prepare the steel for a better zinc coating. When steel members that are fabricated with narrow gaps between overlapping surfaces are subjected to the galvanizing process, the quality of the zinc coating is often inferior along the interface between the two surfaces. This problem occurs because the cleaning solutions—which have a lower viscosity than zinc—get trapped inside the gap while the molten zinc cannot galvanize narrow gaps due to its high viscosity. For this reason, overlapping steel members are seal-welded to prevent such issues. Since seal-welding is not efficient from a fabrication standpoint, the purpose of this research is to provide alternative geometrical overlapping surface configurations. These configurations were subjected to the galvanizing process. To evaluate their performance, three separate tasks were applied through this investigation. The first task involved sealing the gap with commercial silicone sealants instead of a seal-weld, the second task was an accelerated corrosion test, and finally a bond investigation of the coating of the overlapping surfaces was performed. Results have shown that the proposed alternatives did not provide an improved detail for overlapping steel surfaces compared to the existing seal-weld.

Dans les domaines de l'optique et de l'électronique, la fabrication de structures nano et microscopiques organisées nécessitent l'utilisation de méthodes coûteuses d'un point de vue énergétique mais aussi de temps de travail et de quantité de matériel : dans la plupart des cas, les structures sont fabriquées par le biais de photo-lithographie, de gravure, et de FIB (sonde ionique focalisée) sur des matériaux tels que le silicium. Ces méthodes, bien que très utilisées et maîtrisées, produisent des structures dont la surface possède une certaine rugosité causée par le processus de gravure. Une alternative, plus simple et moins coûteuse, se trouve dans l'exploitation du phénomène de démouillage qui peut se définir comme étant une diffusion de la matière gouvernée par des forces interfaciales tentant de réduire la surface entre deux matériaux. En plus des effets décrits ci-après, le démouillage permet d'obtenir une réduction de la rugosité de la surface ; propriété très importante pour une application optique. Le phénomène de démouillage est bien connu dans la littérature et particulièrement dans le domaine des recouvrements polymériques (peintures, traitements de surface, etc.) puisqu'il s'agit d'un e et néfaste causant une réduction de la surface de la couche polymérique appliquée sur un matériau (décollement, formation de gouttelettes, etc.). Mais lorsque ce phénomène est contrôlé et appliqué aux couches-minces, il peut s'avérer très intéressant pour la fabrication de structures à l'échelle microscopique voire nanoscopique. De nos jours il existe plusieurs exemples de couches-minces polymériques ayant subit un démouillage contrôlé pour des applications telles que le développement de surfaces anti-réflectives. De même, le démouillage a été appliqué à des couches-minces métalliques pour former des structures possédant des propriétés plasmoniques (Au, Ag) ou magnétiques (FePt). L'application du démouillage aux systèmes vitreux reste cependant peu développée, mais possède d'attrayantes propriétés, particulièrement pour les verres de chalcogénures. Dans cette thèse, nous discutons des différents types de démouillages appliqués aux couches minces de verres de chalcogénures, et plus particulièrement au système Agx(As20S80)100-x. L'avantage de ce système réside dans sa bonne transmission dans l'infrarouge proche et moyen, permettant ainsi de possibles applications dans cette région sous formes de guides d'ondes, résonateurs, etc. Deux méthodes de démouillages sont étudiées : la première, thermique ; la deuxième, photoII induite. Pour le démouillage induit thermiquement, les couches-minces doivent subir un traitement thermique à des températures supérieures à la température de transition vitreuse du verre a n d'obtenir une viscosité suffisamment faible pour que la tension interfaciale entre la couche-mince et le substrat cause le démouillage de l'échantillon. Cette méthode permet l'obtention de structures circulaires quasiment hémisphériques dont la taille et la forme sont contrôlables par changement de la composition chimique et des paramètres thermiques. Pour le démouillage photo-induit, nous avons utilisé différents lasers à ondes entretenues (CW) dont les longueurs d'onde d'émission sont proches de celles des bandes interdites des différentes couches-minces. Le démouillage ainsi causé par l'irradiation laser est différent du démouillage induit thermiquement. L'exposition au faisceau laser va créer des structures démouillées linéairement et de manière perpendiculaire au champs électrique du faisceau incident. En plus du phénomène de démouillage, lorsque la couche-mince est irradiée par un faisceau laser de longueur d'onde proche de celle de la bande interdite, il apparaît un processus photoanisotropique de biréfringence. Nous avons étudié l'impact des différents paramètres liés au faisceau laser tels que la densité de puissance et le temps d'exposition. Ce processus est indépendant du démouillage et va apparaître avant celui-ci. / The fabrication of self-assembled nano and microscopic structures has been of great interest in the past decades, mainly for the production of optical devices. The methods of fabrication are based on photo-lithography, liquid and/or gas etching, and focused ion beam (FIB). Those methods are costly and time consuming and might create poor surface quality. In this thesis, we present an alternative method of fabrication based on the dewetting phenomenon. It is de ned as the di usion of matter driven by interfacial forces and a reduction of the shared area between two materials. One interesting property of this process is a decrease in surface roughness leading to good optical quality. The dewetting phenomenon is well reported for polymeric lms as well as metallic thin- lms. In the case of polymeric materials, dewetting can be used for the formation of nano and micro-structures, for example, the fabrication of anti-re ective surfaces. For metallic materials, plasmonic (Au, Ag) and magnetic (FePt) structures can be produced on the surface of metallic thin- lms. However, for glassy thin- lms, dewetting is still in its early stages. In this thesis, we will explore di erent types of dewetting on chalcogenide glassy thin- lms of the Agx(As20S80)100-x system. An interesting property of the prepared thin- lms is their transparency in the infrared region allowing promising applications such as waveguides and resonators. Two distinct approaches are studied in this thesis: the rst one is a thermally-induced process; the second one is based on photo-induced e ects. For thermal dewetting, glassy thin- lms must undergo thermal treatments above the glass transition temperature (Tg) in order to decrease the viscosity until interfacial tensions between the lm and its substrate are high enough to induce dewetting. With this technique, we can obtain quasi-hemispherical structures whose shape and size can be controlled by varying chemical composition and thermal parameters (temperature, time, atmosphere, etc.). For laser-induced dewetting, we use a continuous wave (CW) laser whose wavelength must be near the bandgap of the material. Laser-induced dewetted structures are very di erent from their thermal counterparts: instead of being worm-like or hemispherical, they auto-organise in parallel lines, separated by a distance proportional to the initial thickness of the lm. IV Furthermore, structures are always perpendicular to the electric eld of the incident laser beam. Additionally, when thin- lms are exposed to a laser irradiation of near-bandgap energy, a photo-anisotropic process of birefringence appears. We have reported the e ect of di erent laser parameters such as power density and exposure time. This process of photo-birefringence is independant and appears before dewetting.

Next-generation wireless technologies are being actively researched to meet the growing demands for higher data rates, massive connectivity, enhanced reliability, and extended coverage. Recently, reconfigurable intelligent surfaces (RISs) and extremely large antenna arrays (ELAAs) have garnered significant attention as new physical-layer transmission technologies capable of achieving unprecedented spectral and energy efficiency gains. Consequently, RIS and ELAA are considered as promising key enabling technologies for the sixth-generation (6G) and future wireless standards.This dissertation investigates RIS and simultaneously transmitting and reflecting (STAR)-RIS assisted wireless communications, emphasizing design, optimization, and analysis across various practical settings. It presents wireless channel modelling techniques, system design aspects, fundamental performance limits/metrics, including outage probability, average achievable rate, average symbol error rate (SER), diversity order, computational complexity, and algorithmic foundations. This doctoral research also develops algorithms for optimizing RIS/STAR-RIS phase shifts and transmit power allocation in multi-user massive multiple-input multiple-output (MIMO) systems. Moreover, this dissertation characterizes unique propagation characteristics of ELAAs, and thereby impacts of visibility regions (VRs) and spatial non-stationarity in extra-large (XL) RIS communication set-ups with XL-massive MIMO base stations (BS) are analyzed.The dissertation begins with a fundamental performance analysis of RIS-assisted systems operating over Nakagami-m fading channels. It quantifies optimal phase-shifts to maximize received signal-to-noise ration (SNR) and derives the probability distribution of the SNR. The findings include closed-form expressions for outage probability, average SER, and achievable rate, demonstrating that these metrics improve as the number of RIS reflective elements increases. The study also reveals that the achievable diversity order scales linearly with the number of passive RIS elements, resulting in significant diversity gains without additional radio-frequency (RF) chains. Further investigation into STAR-RIS systems with discrete phase-shifts highlights the performance under different protocols, such as energy splitting (ES), mode switching (MS), and time splitting (TS), considering both unicast and multicast transmissions. The analysis demonstrates that employing four-bit phase-shift quantization significantly narrows the performance gap between discrete and continuous phase-shifts. Additionally, it is found that the average achievable rate and SER reach saturation levels at high transmit SNRs, influenced by power allocation coefficients at the transmitter. The dissertation also presents an achievable rate analysis and RIS phase-shift optimization for multi-cell RIS-aided massive MIMO, with for imperfect channel state information (CSI), co-channel interference, and spatially correlated fading. A statistical CSI-based transmit power allocation algorithm is proposed, reducing channel estimation overhead and ensuring user fairness. In exploring STAR-RIS aided multi-user massive MIMO systems, statistical CSI-based STAR-RIS phase-shift and transmit power optimization techniques are used to maximize composite channel gains and ensure fair user rates. The study quantifies the impacts of CSI imperfections, residual interference, and spatially correlated fading. Lastly, the effects of visibility regions in XL RIS setups are examined, deriving achievable user rates and employing phase-shift optimization to maximize user channel covariance. A max-min power allocation algorithm is utilized to address near-far user effects, ensuring system-wide user fairness. Overall, this dissertation provides comprehensive insights and advanced optimization techniques for enhancing RIS and STAR-RIS technologies in wireless communication systems.

No / The rheological behavior of polymeric nanocomposites provides major determination for their processability. In this work, three carbon nanotubes (CNTs) with varied geometries were adopted as nanofiller and then were introduced into poly(methyl methacrylate) (PMMA) matrix with different loadings (0.07–1.0 wt %). The different preparation routine led to varied CNTs dispersion states, on which the shear viscosity and the compressibility of their melts were proved to be sensitive. The technology for the preparation of their nanocomposites played a crucial role in controlling their rheological behavior. With melting blended bare CNTs, the dynamic shear viscosity of PMMA/CNTs increased with the increase of CNTs content, accompanied by aggregated CNTs in which no polymer matrix was entrapped. With the help of surface modification and pre‐mixing, well dispersed CNTs were obtained and a rather low aggregation rate ca. 0.029% was revealed. The well dispersed CNTs with an organic layer which was constructed by small molecules and presented lower viscosity. Such CNTs led to no remarkable clusters within polymer host and played the role of lubricant with an increased‐mobility layer, which can be reflected from the weighted relaxation time spectra.

Ce mémoire présente les démarches qui ont menées à l’obtention d’un modèle statistique pouvant prédire les propriétés d’une surface de fluoropolymère traitée en défilant entre les électrodes d’un système de décharge à barrière diélectrique (DBD) en caractérisant le procédé plasma utilisé. Le réacteur plasma utilisé ainsi que les différents moyens de caractérisation du procédé et de la surface y sont présentés. De plus, plusieurs voies de modélisation utilisant l’algorithme de régression par projection sur les structures latentes (PLS) sont présentées. Afin de modéliser le procédé, une étude de répétabilité en temps de celui-ci a été effectuée. Cette étude avait pour but de montrer que malgré l’apparition d’une couche de dépôt non désirée à la surface de l’électrode découverte du montage DBD, la caractérisation de la surface traitée restait inchangée. Par la suite, la modélisation par PLS est présentée. Avec celle-ci, il est possible d’identifier les facteurs du procédé ayant le plus d’impacts sur la modification de surface. Ces facteurs sont la nature du film de fluoropolymère utilisé comme substrat, la vitesse de défilement de ce film de polymère entre les électrodes de la DBD, le rapport cyclique du signal électrique servant à entretenir la décharge plasma, et le temps de résidence du gaz porteur et du précurseur dans la décharge. Connaissant ces facteurs principaux, des études de cas particuliers sont présentés. Avec ces derniers, il est montré que le modèle devient de moins en moins efficace plus le dépôt est important. Des effets non-linéaires observés sur différentes propriétés de la surface traitée sont aussi observés. / This master thesis contains the steps that lead to a statistical model able to predict the properties of a treated fluoropolymer surface circulating between the electrodes of a dielectric barrier discharge (DBD) system by characterization of the plasma process. The reactor used as well as different characterization apparatus of the plasma process and surface treatment are described. Moreover, different means of modelization using the projection to latent structure (PLS) algorithm are shown. To be able to model the plasma process, a preliminary study of the process repeatability in time has been made. Results of the study show that despite the apparition of an unwanted layer of deposition on the uncovered electrode during the plasma process, the surface treatment physicochemistry does not change. Subsequently, the plasma process modelization by PLS is shown. Using this technique, it is possible to identify and quantify the importance of the input factors in the model. The important factors that are highlighted are the nature of the fluoropolymer film, the line speed of the polymer film between the electrodes, the duty cycle of the electrical signal used to maintain the plasma discharge, and the carrier and precursor gas residence time in the discharge. Knowing these factors, specific case studies were made to assess the proficiency of the model to do predictions. It was observed that the model becomes less precise when the surface shows bigger change. Non-linear effects were also seen of different surface treatment properties.

Mémoire numérisé par la Direction des bibliothèques de l'Université de Montréal.

Le modèle de surface Soil, Vegetation, and Snow (SVS) a été récemment mis au point par Environnement et Changement climatique Canada (ECCC) à des fins opérationnelles de prévisions météorologique et hydrologique. L’objectif principal de cette étude est d’évaluer la capacité de SVS, en mode hors ligne à l’échelle de points de grille, à simuler différents processus par rapport aux observations in situ. L’étude est divisée en deux parties: (1) une évaluation des processus de surface terrestre se produisant dans des conditions sans neige, et (2) une évaluation des processus d’accumulation et de fonte de la neige. Dans la première partie, les flux d’énergie de surface et la teneur en eau ont été évalués sous des climats arides, méditerranéens et tropicaux pour six sites sélectionnés du réseau FLUXNET ayant entre 4 et 12 ans de données. Dans une seconde partie, les principales caractéristiques de l’enneigement sont examinées pour dix sites bien instrumentés ayant entre 8 et 21 ans de données sous climats alpins, maritimes et taiga du réseau ESM-SnowMIP. Les résultats de la première partie montrent des simulations réalistes de SVS du flux de chaleur latente (NSE = 0.58 en moyenne), du flux de chaleur sensible (NSE = 0.70 en moyenne) et du rayonnement net (NSE = 0.97 en moyenne). Le flux de chaleur dans sol est raisonnablement bien simulé pour les sites arides et un site méditerranéen, et simulé sans succès pour les sites tropicaux. Pour sa part, la teneur en eau de surface a été raisonnablement bien simulée aux sites arides (NSE = 0.30 en moyenne) et méditerranéens (NSE = 0.42 en moyenne) et mal simulée aux sites tropicaux (NSE = -16.05 en moyenne). Les performances du SVS étaient comparables aux simulations du Canadian Land surface Scheme (CLASS) non seulement pour les flux d’énergie et le teneur en eau, mais aussi pour des processus plus spécifiques tels que l’évapotranspiration et le bilan en eau. Les résultats de la deuxième partie montrent que SVS est capable de reproduire de manière réaliste les principales caractéristiques de l’enneigement de ces sites. Sur la base des résultats, une distinction claire peut être faite entre les simulations aux sites ouverts et forestiers. SVS simule adéquatement l’équivalent en eau de la neige, la densité et la hauteur de la neige des sites ouverts (NSE = 0.64, 0.75 et 0.59, respectivement), mais présente des performances plus faibles aux sites forestiers (NSE = - 0.40, 0.15 et 0.56, respectivement), ce qui est principalement attribué aux limites du module de tassement et à l’absence d’un module d’interception de la neige. Les évaluations effectuées au début, au milieu et à la fin de l’hiver ont révélé une tendance à la baisse de la capacité de SVS à simuler SWE, la densité et l’épaisseur de la neige à la fin de l’hiver. Pour les sites ouverts, les températures de la neige en surface sont bien représentées (RMSE = 3.00 _C en moyenne), mais ont montré un biais négatif (PBias = - 1.6 % en moyenne), qui était dû à une mauvaise représentation du bilan énergétique de surface sous conditions stables la nuit. L’albédo a montré une représentation raisonnable (RMSE = 0.07 en moyenne), mais une tendance à surestimer les valeurs de fin d’hiver (biais = 0,04 sur la fin de l’hiver), en raison de la diminution progressive pendant les longues périodes de fonte. Enfin, un test de sensibilité a conduit à des suggestions aux développeurs du modèles. Les tests de sensibilité du processus de fonte de la neige suggèrent l’utilisation de la température de surface de la neige au lieu de la température moyenne lors du calcul. Cela permettrait d’améliorer les simulations SWE, à l’exception de deux sites ouverts et d’un site forestier. Les tests de sensibilité à la partition des précipitations permettent d’identifier une transition linéaire de la température de l’air entre 0 et 1 _C comme le meilleur choix en l’absence de partitions observées ou plus sophistiquées. / The Soil, Vegetation, and Snow (SVS) land surface model has been recently developed at Environment and Climate Change Canada (ECCC) for operational numerical weather prediction (NWP) and hydrological forecasting. The main goal of this study is to evaluate the ability of SVS, in offline point-scale mode, to simulate different processes when compared to in-situ observations. The study is divided in two parts: (1) an evaluation of land-surface processes occuring on snow-free conditions, and (2) and evaluation of the snow accumulation and melting processes. In the first part, surface heat fluxes and soil moisture were evaluated under arid, mediterranean, and tropical climates at six selected sites of the FLUXNET network having between 4 and 12 years of data. In the second part, the main characteristics of the snow cover are examined at ten well-instrumented sites having between 8 and 21 years under alpine, maritime and taiga climates from ESM-SnowMIP network. Results of the first part show SVS’s realistic simulations of latent heat flux (NSE = 0.58 on average), sensible heat flux (NSE = 0.70 on average), and net radiation (NSE = 0.97 on average). Soil heat flux is reasonably well simulated for the arid sites and one mediterranean site, and poorly simulated for the tropical sites. On the other hand, surface soil moisture was reasonably well simulated at the arid (NSE = 0.30 on average) and mediterranean sites (NSE = 0.42 on average) and poorly simulated at the tropical sites (NSE = - 16.05 on average). SVS performance was comparable to simulations of the Canadian Land Surface Scheme (CLASS) not only for energy fluxes and soil moisture, but more specific processes such as evapotranspiration and water balance. Results of the second part show that SVS is able to realistically reproduce the main characteristics of the snow cover at these sites. Based on the results, a clear distinction between simulations at open and forest sites can be made. SVS is able to simulate well snow water equivalent, density and snow depth at open sites (NSE = 0.64, 0.75 and 0.59, respectively), but exhibits lower performances over forest sites (NSE = - 0.40, 0.15 and 0.56, respectively), which is attributed mainly to the limitations of the compaction scheme and the absence of a snow interception scheme. Evaluations over early, mid and end winter periods revealed a tendency to decrease SVS’s ability to simulate SWE, density and snow depth during end winter. At open sites, SVS’ snow surface temperatures are well represented (RMSE = 3.00_C on average), but exhibited a cold bias (PBias = - 1.6% on average), which was due to a poor representation of the surface energy balance under stable conditions at nighttime. Albedo showed a reasonable representation (RMSE = 0.07 on average), but a tendency to overestimate end winter albedo (bias = 0.04 over end winter), due to the slow decreasing rate during long melting periods. Finally, sensitivity tests to the snow melting process suggest the use of surface snow temperature instead of the average temperature when computing the melting rate. This would provide the improvement of the SWE simulations, with exception of two open and one forest sites. Sensitivity tests to partition of precipitation allows to identify a linear transition of air temperature between 0 and 1_C as the best choice in the absence of observed or more sophisticated partitions

Call number: LD2668 .T4 1979 D465 / Master of Science

Thesis (MSc)--Stellenbosch University, 2012. / ENGLISH ABSTRACT: Subdivision is an important and e cient tool for rendering smooth curves and surfaces in computer graphics, by repeatedly applying a subdivision (re ning) scheme to a given set of points. In the literature, attention has been mostly restricted to developing binary subdivision schemes. The primary emphasis of this thesis is on ternary subdivision, and in particular on the interpolatory case. We will derive a symmetric ternary interpolatory subdivision scheme for the rendering of curves, satisfying analogous properties to the Dubuc-Deslauriers binary scheme. Explicit construction methods, as well as a corresponding convergence analysis, will be presented. Graphical illustrations of the results will also be provided. / AFRIKAANSE OPSOMMING: Subdivisie bied 'n belangrike en doeltre ende metode om gladde krommes en oppervlakke in rekenaargra ka te genereer. Hierdie metode behels dat 'n subdivisieskema (of verfyningskema) herhaaldelik toegepas word op 'n gegewe versameling punte. In die literatuur word daar hoofsaaklik gefokus op die ont- wikkeling van bin^ere subdivisieskemas. In hierdie tesis word die klem gel^e op tern^ere subdivisieskemas, en in die besonder op interpolerende skemas. Ons sal 'n simmetriese tern^ere interpolerende subdivisieskema, wat analo e eienskappe as di e van die Dubuc-Deslauriers bin^ere skema bevredig, ontwikkel, om krom- mes te lewer. Eksplisiete konstruksiemetodes en ooreenkomstige konvergensie- analise, asook gra ese illustrasies van die resultate, sal getoon word.