Cam mechanisms are widely used in industry, in applications requiring quick-return and indexing motions. A current research effort at the Robotic Mechanical Systems Laboratory of McGill University's Centre for Intelligent Machines aims at the application of cam mechanisms as speed reducers. The accuracy required in these mechanisms is of the utmost importance, especially when cams are rotating at a high speed. / In this thesis, the design for manufacturability of planar speed-reduction cam mechanisms is studied. In particular, the thesis focuses on a speed reducer with a rotating follower to couple shafts of parallel axes, termed planar Speed-o-Cam. Principles of the design for manufacturability are applied to Speed-o-Cam and a unified method for obtaining the optimum parameters satisfying the curvature constraints and pressure-angle bounds is developed. These two factors are relatively important because Numerically Controlled and Computer Numerically Controlled machine tools could be very sensitive to changes of curvature of the workpiece, especially when milling complex shapes such as those of cam plates. / Cam-mechanism balancing is also studied because unbalance in a high-speed rotating element can cause severe vibrations and greatly affect the bearings and hence, the performance of the machine. This is done by not only adding counterweights, which unavoidably increase the weight and volume of the mechanism, but also by removing material.

The problem of direct initiation of detonations has been investigated numerically by using a more realistic chemical kinetics scheme. The chemical kinetics model is based on a three-step chain-branching reaction consists sequentially of a chain-initiation and a chain-branching step, followed by a temperature independent chain-termination. The steady ZND structure using the 3-step chemical kinetics model is compared with that based on the standard single-step Arrhenius rate model. An unsteady computation has also been performed to determine whether the solution from steady-state analysis can be approached asymptotically from a transient development. / The three regimes of direct initiation have been numerically simulated for planar, cylindrical and spherical geometries using the present 3-step chemical kinetics model. The use of a more realistic reaction scheme allows a unique value for the critical initiation energy to be defined. / Some general theories for direct initiation have been verified from the results of the present numerical simulation using a more detailed chemical kinetics model, where a definitive value of the critical energy was obtained. (Abstract shortened by UMI.)

This thesis reports on the design, analysis and optimization of an accelerometer. This accelerometer is designed as a monolithic structure along the concept of compliant mechanisms. An elastodynamic analysis is performed on the compliant mechanism to asses the performance of the inertial sensor. This thesis proposes an innovative compliant hinge intended to stiffen the struc- ture of compliant mechanisms. In addition, a procedure for the optimum design of this new hinge is discussed. The structural optimization problem is approached by coupling a finite element model to an optimization algorithm. A procedure is devel- oped to generate the mesh at each optimization step according to the values of the design parameters provided by the optimization algorithm. The objective function to minimize is the stress concentration in a hinge loaded under bending. The last chapter focuses on the multi-objective optimization of the compliant- mechanism accelerometer. The Pareto method is used to optimally design the ac- celerometer. The purpose is to maximize the sensitivity of the accelerometer in its sensing direction, while minimizing its sensitivity in all other directions. The a poste- riori multi-objective optimization is formulated. By using the normalized normal constrained method (NNCM), an even distribution of the Pareto frontier is found. The work also provides several optimum solutions of the Pareto plot as well as the CAD model of the selected solution. / Une large gamme d'accéléromètres est offerte sur le marché. Cependant, la plupart des architectures de capteurs inertiels offerts par l'industrie sont constituées d'une masse suspendue par une poutre encastrée. Avec les années, les chercheurs ont mis au point des architectures parallèles offrant une bien meilleure rigidité qu'une simple poutre encastrée. Les accéléromètres à architecture parallèle offrent également de bien meilleures rigidités. Cette thèse porte sur la conception, l'analyse et l'optimisation d'un accéléromètre à architecture parallèle. Tout d'abord, l'accéléromètre est réalisé comme structure monolithique dans le cadre des mécanismes flexibles. Par la suite, une analyse elasto-dynamique est effectuée sur le mécanisme flexible afin de d'´evaluer les performances du capteur inertiel. Cette thèse propose également une nouvelle articulation flexible visant à améliorer la structure des mécanismes flexibles. Une procédure optimisant le profil de cette nouvelle articulation flexible est également proposée. Le problème d'optimisation structurelle est abordé en établissant une boucle entre un modèle par éléments finis et un algorithme d'optimisation. Une procédure a été développée afin de générer le maillage à chaque étape d'optimisation en fonction des valeurs des paramètres de conception fournis par l'algorithme d'optimisation. La fonction cible à minimiser est définie comme la concentration de contrainte générée dans l'articulation flexible sollicitée en flexion. Le dernier chapitre de la thèse met l'accent sur l'optimisation multi-objectif du mécanisme flexible de l'accéléromètre. La méthode

Aqua is an underwater hexapod robot that uses paddles to propel and orient itself. The system is highly non-linear and coupled, and thus far a controller has not been implemented on the robot. In this work, three different controllers were developed and utilized on the robot. The design of a controller for the vehicle began with the development of a stability augmentation system (SAS). In order to study the stability of the system, the model needed to be linearized and this was accomplished using numerical differentiation by finite differences. Using this method state space matrices were derived for three different steady state velocities and corresponding SAS's were designed based on the system's eigenvalues. These SAS's were implemented in a non-linear simulation and were shown to need further refinement. The refined SAS's were then designed and were successfully implemented on the physical robot in fresh and sea water. The design of an autopilot to operate with the SAS followed which included a proportional and a proportional-integral controller. The controllers were tested in simulation and in experiment with inconsistent results. Finally, the SAS was modified to compensate for possible faults that may occur during the operation of the robot. It was found that the original SAS was sufficiently robust to compensate for the case of a missing flipper. However, the case of a flipper stuck at a fixed angle required a modification to the SAS and this was accomplished by analyzing the additional drag forces created by the fault. The modified SAS was implemented on the robot in a set of experiments with successful results. / Aqua est un robot hexapode sous-marin utilisant des palmes comme moyen de propulsion et de direction. Ce robot forme un système non linéaire et couplé, et présentement il n'y a pas un contrôleur sur le robot. Dans cette oeuvre, trois contrôleurs différents ont été développé et implanté sur le robot. La conception du contrôleur a débuté par le développement d'un système d'augmentation de stabilité (SAS). Pour étudier la stabilité du système, le modèle a été linéarisé, ceci par une méthode de différences finis. Avec cette méthode, les matrices du système ont été obtenues pour trois vitesses différentes et les SAS's ont été conçu a partir des valeurs propres de système. Ces SAS's ont été utilisés dans les simulations non linéaire et ils ont dû être ajustés. Ces SAS's ont été implantés sur le robot en eau douce et salée. Puis un contrôleur de pilote automatique a été conçu pour fonctionner avec le SAS, incluant des contrôleurs proportionnels et proportionnel-intégrales. Les contrôleurs ont été utilisés en simulation et en expérience avec les résultats variables. Finalement, le SAS a été modifié pour compenser les fautes possibles qui peuvent se passer pendent l'utilisation du robot. On a trouvé que le SAS original était suffisamment robuste pour compenser pour la faute d'une palme manquante. Mais, pour le cas d'une palme coincée à un angle fixe, il a fallu modifier le SAS. Ceci a été fait à partir d'une analyse des forces additionnelles créées par la faute. Le SAS modifié a été mis en pratique sur le robot avec des résultats positifs.

This thesis addresses the question: can an electric snowmobile be one of the solutions to help lower snowmobile emissions and energy consumption? In addressing this question the performance limitations of current electric snowmobile prototypes are investigated and it is shown that, unless a huge leap is seen in current battery technology energy density, electric snowmobiles cannot perform on par with gasoline snowmobile on both range and performance simultaneously. Despite this, electric snowmobiles do have a certain number of niche applications where they can be useful. This thesis suggests that electric snowmobile powetrain modeling and simulation for these niche applications can potentially help overcome some of the challenges that exist in implanting such a vehicle for regular use. A complete, virtual electric snowmobile model was built and validated using actual electric snowmobile on-snow test data. Snowmobile emission and energy consumption simulation was performed and demonstrated that Canadian electric snowmobile fuel cycle emissions and energy consumptions were, in general, substantially lower than gasoline snowmobiles. However, this is closely linked with electricity generation techniques and should not be extrapolated to say that this is the case for all potential electric snowmobiles worldwide. / Cette thèse tente d'apporter réponse à la question suivante: est-ce que la motoneige électrique peut faire partie des solutions afin d'aider à diminuer les émissions et la consommation d'énergie des motoneiges? Afin de répondre à cette question, un questionnement sur les performances des prototypes de motoneiges électriques actuelles est entrepris et il en ressort qu'à moins qu'une énorme percée technologique ne vienne changer la donne, la densité d'énergie des technologies de piles présentement disponibles fait en sorte qu'une motoneige électrique ne peux pas envisager performer de façon similaire à une motoneige conventionnelle à essence sur le plan de l'autonomie et de la performance de façon simultanée. Ceci étant dit, il existe tout de même des applications de niche pour lesquelles une motoneige électrique est parfaitement apte à accomplir le travail requis. Cette thèse suggère que l'utilisation de la modélisation et de la simulation peut s'avérer un outil précieux afin de surmonter certains obstacles à l'implantation de motoneiges électriques. Un modèle virtuel de motoneige électrique est assemblé et une simulation virtuelle est complétée et ensuite validée en utilisant des données obtenues lors d'essais sur neige avec une vraie motoneige électrique. Une simulation est faite et permet de démontrer que sur la totalité de son cycle de carburant, une motoneige électrique canadienne émet généralement moins d'émissions et consomme moins d'énergie qu'une motoneige conventionnelle à essence, et ce, de façon substantielle. Cependant, puisque ce résultat est fortement lié aux méthodes de production d'énergie il ne peu

It is common practice to add reactive metal particles to high-explosive formulations in order to increase the total energy output. The present study is an effort to increase the understanding of metal particle combustion in detonation products. Aluminum and magnesium particles ranging from 2 to 100 μm are subjected to the flow of detonation products of a stoichiometric mixture of hydrogen and oxygen. Luminosity emitted from the reacting particles is used to determine the reaction delay and duration. It is found that the reaction duration increases as d^n with n ≈ 0.5, more consistent with kinetically controlled reaction rather than the classical diffusion controlled regime. Emission spectroscopy is used to estimate the combustion temperature, which is found to be well below the flow temperature. This fact also suggests combustion in the kinetic regime. Finally, the flow field is modelled with a CFD code and the results are used to model analytically the behaviour of the aluminum particles. Furthermore, a simple analytical model is developed to predict ignition of magnesium particles in nitromethane detonation products. The flow field is simplified by considering the detonation products as a perfect gas expanding in a vacuum in a planar geometry. This simplification allows the flow field tobe solved analytically. A single particle is then introduced in this flow field. Its trajectory and heating history are computed. It is found that most of the particle heating occurs in the Taylor wave and in the quiescent flow region behind it, shortly after which the particle cools. By considering only these regions, thereby considerably simplifying the problem, the flow field ca / Il est maintenant courant d'ajouter des particules de métaux réactifs aux formulations d'explosifs pour augmenter l'énergie libérée. Cette étude vise une meilleure compréhension de la combustion de métaux dans les produits de détonation. Des particules d'aluminium et de magnésium de 2 à 100 μm sont exposées à l'écoulement des produits de la détonation d'un mélange d'hydrogène et d'oxygène. La lumière émise par les particules est utilisée pour déterminer le délai et la durée de la réaction. La durée de la réaction augmente selon d^n avec n ≈ 0.5. Ces résultats suggèrent un régime de combustion cinétique plutôt que le régime diffusif classique. La température de combustion est mesurée à l'aide de spectroscopie d'émission. La température mesurée est plus basse que celle de l'écoulement, ce qui est également cohérent avec un régime de combustion cinétique. Finalement, l'écoulement est modélisé avec un code Euler, dont les résultats sont utilisés pour modéliser analytiquement le comportement des particules. Dans un deuxième temps, un modèle analytique simple est développé pour prédire l'ignition de particules de magnésium dans les produits de détonation de nitrométhane. L'écoulement est simplifié en traitant les produits comme gaz parfaits, qui prennent de l'expansion dans le vide dans une géométrie planaire. Ces simplifications permettent de résoudre l'écoulement analytiquement. Ensuite, la trajectoire et la température d'une seule particule introduite dans l'écoulement sont calculées. La particule est chauffée principalement dans le "Taylor wave" et dans la région quiescente derrière celle-ci. En ne

This thesis presents a detailed kinematic analysis of three degree-of-freedom planar parallel manipulator platforms possessing topological symmetry, called general planar Stewart-Gough platforms (PSGP). A specific super-set of topologically asymmetric platforms and one with actuated holonomic higher pairs are included in the analysis. / After PSGP are described and classified, the remainder of the first portion is devoted to the review of the geometric and mathematical tools used in the analysis. / A single univariate polynomial is derived which yields the solutions to the forward kinematics problem of every PSGP platform. Kinematic mapping is used to represent distinct displacements of the platform as discrete points in a three-dimensional projective image space. Separate motions of each leg map to skew one-sheet hyper-boloids, or hyperbolic paraboloids, depending on the kinematic architecture of the leg. After two elimination steps the three quadric surfaces are reduced to a sixth order univariate. The roots of this polynomial reveal all solutions to the forward kinematics problem. The procedure leads to a robust algorithm which can be applied to the abovementioned super-set. / The inverse kinematics problem of these platforms is solved, in closed form, using the same kinematic mapping. The procedure can be applied to any three-legged planar platform with lower pairs, regardless of symmetry. / A workspace analysis and simple criteria for the determination of the existence of a dextrous workspace are presented. Finally, a geometric singularity and self-motion detection method, which does not employ Jacobian matrices, is discussed.

Loosely supported cylinders subjected to cross-flow may undergo fluidelastic instability in the support inactive mode resulting in cylinder/support impacting. The cylinder/support interaction forces and, in turn, the resulting cylinder wear rates are strongly dependent on the detailed dynamical response. This Thesis examines the response of a loosely supported cylinder located in the third row of an otherwise rigid rotated triangular array. The feasibility and potential of a modern nonlinear dynamics approach to the understanding of the underlying dynamics is investigated. / A nonlinear quasi-steady model was formulated to model the dynamical behaviour. The steady fluid force field, required as input to the model, was measured experimentally for a cylinder within a rotated triangular array. A lither stability analysis showed the cylinder stability behaviour to be strongly dependent on cylinder position. This result serves as a possible explanation for the rare occurrence of, theoretically predicted, multiple instability regions in experimental measurements. / The nonlinear analysis uncovered two important transition routes to chaos. The first, a switching mechanism prevalent at the onset of impacting. The second and most important is the intermittency route to chaos. The theoretical model showed good agreement with experiments in predicting the bifurcation sequences and transitions to chaos--comparisons were quantified via fractal dimensions and saddle orbit distributions. / The identification of type I intermittency leads to a quantitative estimate of the probability distribution of the length of laminar phases. It is shown that the duration of laminar phases and the associated frequency may provide better estimates of integration time and frequency for wear-rate computation.

This thesis deals with the nonlinear dynamics of an articulated system of cylinders in confined axial flow. The articulated cantilevered system is composed of rigid cylindrical segments, interconnected by rotational springs, and is hanging vertically in the centre of a cylindrical pipe, with fluid flowing downwards in the narrow annular passage. / The equations of motion were obtained by application of Lagrange's equations, for a system with an arbitrary number of articulations. The forces associated with the structure itself, i.e. the inertial, restoring and gravity forces acting on the structure, are taken into account in the kinetic and potential energies of the system. The hydrodynamic forces are incorporated partly in the kinetic energy and partly as generalized forces. / As the articulated system interacts with the outer pipe, this interaction or impact is modelled by a cubic or a trilinear spring, or by using the coefficient of restitution method. / The critical flow velocity for the onset of fluidelastic instabilities, such as divergence or flutter, is calculated by a linear eigenvalue analysis. Then, two models for the equations of motion are investigated, both analytically and numerically. Centre manifold and normal form theory are used to calculate the post-Hopf limit cycle amplitude, which will be compared with that obtained numerically. / Phase portraits, power spectral densities and bifurcation diagrams indicate in some cases a clear period-doubling cascade leading to chaos, while in others chaos arises via the quasiperiodic route or via type III intermittency. In addition, Poincare maps and Lyapunov exponent calculations confirm the existence of quasiperiodicity or chaotic motion.

The formulation of a co-located equal-order Control-Volume-based Finite Element Method (CVFEM) for the solution of two-fluid models of two-dimensional, planar or axisymmetric, incompressible, gas-solid particle flows is presented in this thesis. The main focus is on the development of a numerical method that allows computer simulation of gas-solid particle flows over a wide range of solid-phase volume concentration in complex irregular geometries. / A general two-fluid mathematical model is presented. This model is essentially borrowed from published works in the area of granular flows. It is established here that this model is applicable to gas-solid flows over a wide range of solid-phase concentration. The governing equations of the fluid phase are obtained by volume averaging the Navier-Stokes equations for an incompressible fluid. The solid-phase macroscopic equations are derived using an approach that has been successfully used earlier for the description of granular materials, and is based on the kinetic theory of dense gases. This approach accounts for particle/particle collisions, and permits the determination of the solid-phase macroscopic properties such as viscosity and pressure. / The proposed CVFEM is formulated by borrowing and extending ideas put forward in earlier CVFEMs for single-phase flows. In axisymmetric problems, the calculation domain is discretized into torus-shaped elements and control volumes: in a longitudinal cross-sectional plane, or in planar problems, these elements are three-node triangles, and the control volumes are polygons obtained by joining the centroids of the three-node triangles to the midpoints of the sides. In each element, mass-weighted skew upwind functions are used to interpolate the volume concentrations. An iterative variable adjustment algorithm is used to solve the discretized equations. / The chosen mathematical model, along with its specializations to single-phase flows and dilute gas-solid flows, and the proposed CVFEM have been applied to several test problems and some demonstration problems. These test and demonstration problems include single-phase flows, dilute-concentration gas-solid particle flows and dense-concentration gas-solid particle flows. The CVFEM results have been compared with results of independent numerical and experimental investigations whenever possible. These comparisons and the results of the demonstration problems are quite encouraging.