Numerical simulation of forming processes has been an important means for material selection, tool design, and process optimization. A critical component of simulation, however, is an accurate material constitutive model, describing the response of the material under possible modes of deformation. The accuracy, in turn, is linked to the tests and techniques applied for identification of constitutive models: the more elaborate the identification, the more reliable the material parameters. For textile composites, uncontrollable factors such as contact friction, misalignment, slip, variations in local fiber volume, and tow compaction are sources that generate considerable scatter in the response of fabrics. Accordingly, characterization methods occasionally suffer from non-repeatability of test data even under similar testing conditions. Furthermore, it is typical that different deformation modes result in different sets of material parameters. If variance of material response within the replication of tests and deformation modes is neglected, then the identification of model parameters can be far from the true material behavior. In order to confront the above shortcomings, this work is an attempt to elaborate on the characterization of textile composites using a new inverse method by means of a signal-to-noise weighting scheme, and two constitutive models by means of a phenomenological invariant-based approach. A full identification of the developed constitutive models for a typical woven fabric is applied using the introduced inverse method and a set of data from standard testing methods, with close attention to the behavior of the composite constituents in a macro level. Particularly, the effects of fiber-resin interactions and fiber misalignment are introduced. A novel modified picture frame test is also studied and used for validating the models. From the results of this work, it is expected that the use of a number of test methods simultaneously and the inclu / La simulation numérique pour les procédés de formation est un instrument important pour le choix des matériaux, la conception d'outils et l'optimisation des procédés. Un des composants critiques de la simulation reste cependant un modèle constitutif de matériaux précis, pouvant décrire la réponse du matériau sous des modes de déformation possibles. Par ailleurs, l'exactitude est liée aux essais et aux techniques appliqués pour l'identification des modèles constitutifs: plus l'identification est raffinée, plus les paramètres du matériau sont fiables. Pour des composés textiles, des facteurs incontrôlables tels que le frottement de contact, la déviation d'alignement, le glissement, les variations de volume local des fibres, et le tassement des fibres sont des sources qui produisent un éparpillement considérable dans la réponse des tissus. Par conséquent, les méthodes de caractérisation souffrent occasionnellement de la non-répétabilité des données des essais même lorsque les conditions d'essais sont semblables. De plus, il est typique que différents ensembles de paramètres du matériau soient obtenus à partir des différents modes de déformation. Si la variance de la réponse du matériau dans les répliques d’essais et les modes de déformation est négligée, l'identification des paramètres du modèle peut alors être loin du véritable comportement du matériau. Afin de confronter les imperfections mentionnées ci-dessus, la présente étude est une tentative d'élaboration de la caractérisation des composés textiles en utilisant une nouvelle méthode inverse basée sur un schéma pondéré signal/bruit et deux modèles constitutifs par le biais d’une approche basée sur l’invariance des phénomènes. Une identification complète des modèles constitutifs développés est appliquée à un tissu typique en utilisant la méthode inverse proposée et un ensemble de données obtenus à partir des méthodes d'essais stand

This thesis reports work not only on the numerics behind, but also the implementation of real-time redundancy-resolution schemes on a real robot. Furthermore, the robot that was chosen as the experimental platform is, isotropic. Redundant robots have more degrees of freedom than needed to perform a class of tasks. A new performance index is proposed in the thesis, that is quadratic in the joint variables and its weighting function has units of frequency. It is shown how this performance index can produce cyclic trajectories in a simple manner, thereby eliminating the undesired drift of the joint angles upon tracking a closed Cartesian trajectory. / Isotropic robots can be postured in such a way that the condition number of their Jacobian matrices can attain a minimum value of unity. It is shown in the thesis that this feature is closely related to the performance of the robot. It appears that trajectories that are close to the set of isotropic postures are performed with smaller errors than trajectories lying far from that set. / The experiments reported here were conducted on the McGill-IRIS C3 Arm, an isotropic, four-revolute redundant manipulator, used to position the operation point of its end link. (Abstract shortened by UMI.)

The thermal expansion behavior of Duralcan particle reinforced composite materials was investigated. Initially, the temperature dependence of the CTE of Al-Si alloy containing SiC reinforcement particles ranging from 10 to 40% in volume was experimentally examined and compared with standard theoretical model predictions. In addition, the effects of reinforcement volume fraction and nature of the composite constituents during thermal cycling between 25 and 350ºC were determined for Al-Si alloy containing between 10 and 40% SiC particles and Al alloy having 40% alumina in volume. Accurate experimental CTE measurements were made using a high precision Thermomechanical Analyzer system. Silicon carbide reinforced composite average CTE values were bounded by two elastic CTE theoretical models consisting of Schapery and Kerner predictions over the 25--350ºC interval for reinforcement volume fractions between 10 and 40%. The CTE mismatch between the particles and the matrix does not appear to be the only factor influencing the expansion response of the composites. Indeed, the nature of the composite constituents also plays an important role by influencing the ductility and bonding of the particle-matrix interface.

An unstructured-grid, h-adaptive control-volume finite element method (CVFEM) was formulated, implemented, and tested for the simulation of steady, viscous, incompressible fluid flow and heat transfer in arbitrarily shaped two-dimensional domains. The h-adaptive procedure is based on a posteriori error estimation using a superconvergent recovery or smoothing technique. The CVFEM is based on a co-located, equal-order formulation that deals directly with the primitive variables. A set of sufficient conditions is presented for guaranteeing that the algebraic approximations to the diffusion terms in the integral conservation equations contribute positively to the coefficients in the discretized equations. An additive-correction multigrid method, based on an adaptive volume-agglomeration technique, was implemented for the purposes of accelerating iterative convergence in the solution of the discretized equations. / An extended Richardson extrapolation technique for unstructured grids is proposed: A discrete solution is obtained from an h-adaptive analysis in which the estimated error is reduced to a predetermined level and uniformly distributed over each element. The resulting mesh is then used as a base grid in constructing a hierarchy of grids via uniform element subdivision. / A closed-form, analytical solution to the steady, two-dimensional, incompressible Navier-Stokes equations was obtained by prescribing a divergence-free velocity field and a consistent pressure field, substituting them into the momentum equations, and deriving the implied volumetric source terms. This analytical solution was used to test several numerical algorithms implemented in this work. Lid-driven, viscous flow in a triangular cavity was also used as a test problem.

We have built a planar one-legged robot (the ARL Monopod) to study the design, control and energetics of autonomous dynamically stable legged machines. Our 15kg robot is actuated by two low power 80W DC electric motors, yet it operates in a stable and robust fashion up to a running speed of 4.3km/hr (1.2m/s). Both design and control of our machine borrow heavily from Raibert's work whose robots use much more powerful hydraulic actuators. Despite our comparatively low power available for actuation, only the hopping height controller had to be modified to achieve stable running. In order to improve energetic efficiency we introduced a scalar "locomotion time" variable, which maps one locomotion cycle onto a fixed interval, independent of operating conditions. This locomotion time has allowed the implementation of a reference trajectory for flight phase leg angle tracking. We have performed a detailed energetic analysis of robot running for the ARL monopod in order to determine how its efficiency may be improved. A comparative energetic analysis reveals our machine as the most energy efficient powered legged robot having a specific resistance of $ varepsilon=0.7$ at 4.3km/hr (1.2m/s) running speed.

An experimental investigation of the influence of upstream turbulence on discrete-hole film cooling of a model blade in confined cross flow is presented and discussed. A model blade and test section were designed and constructed. A system for constant-temperature hot-wire anemometry was implemented. A microcomputer-based data acquisition, control, and processing system was designed and implemented. The model blade was a blunt body with a semicircular leading edge, a flat after body, and a tapered trailing edge. Local heat transfer coefficients without injection of a secondary fluid were determined using a thin-film technique. In the film cooling studies, two rows of discrete injection holes at $ pm$30$ sp circ$ from the stagnation line were used. The injection tubes were oriented normal to the surface of the model blade and were coplanar with the primary flow velocity vector. / Nominal turbulence intensities of 0.7%, 9.8%, and 14.4% at a location 1.3 diameters upstream of the stagnation line on the semicircular leading edge of the model blade were investigated. The range of Reynolds number, based on the diameter of the semicircular leading edge and upstream velocity, was 23,000 to 75,000. In the discrete-hole film cooling studies, mass flux ratios in the range 0.6 $<$ M $<$ 2.0 were considered. The heat transfer results are presented in terms of the distributions of local Nusselt number and a nondimensional temperature on the surface of the blade. The results of the film cooling studies are presented in terms of the distributions of the effectiveness.

A computational and experimental study was made of the steady developing laminar convective heat transfer to viscous non-Newtonian fluids described by the power law model flowing in straight channels of circular and several non-circular cross-sections. In the numerical study the governing conservation equations in three dimensions subject to suitable boundary conditions were solved after appropriate discretization, using the Galerkin finite element method. Fourteen different cross-sectional geometries were studied numerically. Effects of temperature-dependent apparent viscosity, viscous dissipation as well as Prandtl number were included in the model. The thermal boundary conditions tested were: uniform wall temperature and uniform wall heat flux on the entire duct surface. Predicted velocity fields, pressure drop, and heat transfer distributions were compared with available data and simulation results as appropriate. Results are presented on the comparative thermal performance of various cross-sectional ducts. / In the experimental study the critical Reynolds numbers were measured for distilled water flowing through a semi-circular and an equilateral triangular duct. Local Nusselt number distributions are presented for ducts with the horizontal bottom plane sides subjected to uniform heat flux while the rest of the channel is well insulated. The measured pressure drop and Nusselt number distribution for Carbopol solutions compared very well with the numerical predictions.

This thesis is devoted to the kinematic synthesis of parallel manipulators at large, special attention being given to three versions of a novel class of mainpulators, named double-triangular. These are conceived in planar, spherical and spatial double-triangular varieties. / The treatment of planar and spherical manipulators needs only planar and spherical trigonometry, a fact that inductively leads to the successful treatment of spatial varieties with methods of spatial trigonometry, wherein the relationships are cast in the form of dual-number algebraic expressions. Using the foregoing tools, the direct kinematics of the three types of double-triangular manipulators is formulated and resolved. / Moreover, a general three-group classification, to deal with singularities encountered in parallel manipulators, is proposed. The classification scheme relies on the properties of Jacobian matrices of parallel manipulators. It is shown that all singularities, within the workspaces of the manipulators of interest, are readily identified if their Jacobian matrices are formulated in an invariant form. / Finally, the optimal design of the manipulators is studied. These designs minimize the roundoff-error amplification effects due to the numerical inversion of the underlying Jacobian matrices. Such designs are called isotropic. Based on this concept the multi-dimensional isotropic design continua of several manipulators are derived.

In this thesis, the dynamics of quadrupedal running via the bounding gait is studied. To analyse the properties of the passive dynamics of Scout II, a model consisting of a body and two massless spring-loaded prismatic legs is introduced. A return map is derived to study the existence of periodic system motions. Numerical studies of the return map show that passive generation of cyclic motion is possible. Most strikingly, local stability analysis of the return map shows that the dynamics of the open loop passive system alone can confer stability of the motion. Stability improves at higher speeds, a fact which is in agreement with recent results from biomechanics showing that the dynamics of the body become dominant in determining stability when animals run at high speeds. Furthermore, pronking is found to be more unstable than bounding, which explains why Scout II shows a "preference" for the bounding gait. These results can be used in developing a general control methodology for legged robots, resulting from the synthesis of feed-forward and feedback models that take advantage of the mechanical system.

The rate of heat transfer to a fluid flowing in a pipe can be greatly enhanced by the use of internal fins, but the pressure gradient must be similarly increased in order to maintain the desired flow rate. Computer simulations are a cost-effective method to obtain optimal heat transfer designs of internally finned pipes. / In this work, a control-volume finite element method was formulated, coded, and used for computer simulations of fully-developed, laminar, fluid flow and heat transfer in pipes with internal twisted fins of triangular cross-section. Values of Nusselt number and friction factor were computed for a relevant range of twist ratio, fin length, fin thickness, and Reynolds number, for a Prandtl number of 0.7. Some optimal configurations were identified. An overview of this work and the results will be presented in this thesis.