Many a mechanical application involves power transmission from a high-speed motor to a low-speed load. However, existing speed-reduction mechanisms are usually a major sink of energy and information in mechanical transmissions. Energy and positioning information are lost through: (a) friction between sliding components; (b) compliance; and (c) backlash. A novel transmission for speed reduction, Speed-o-Cam, is currently under research at McGill University's Centre for Intelligent Machines (CIM). The transmission is based on the layout of pure-rolling indexing cam mechanisms, and hence, eliminates backlash and friction. Besides zero backlash and low friction losses, Speed-o-Cam also offers the possibility of high stiffness, another essential attribute for high-accuracy applications. / This thesis focuses on the aspects of both model development and mechanical-parameter identification of a spherical prototype of Speed-o-Cam. Our main interest lies in identifying the mechanism stiffness. In order to conduct experiments on the prototype, a testbed was designed and fabricated. A mathematical model of the testbed is first formulated. Based on this model and the results of experiments, the parameters of the Speed-o-Cam prototype are identified. In the process, the stiffness and damping parameters of the couplings of the testbed are also identified. / Power efficiency is an important indicator of speed reducing mechanisms. For the Speed-o-Cam prototype, this indicator is also estimated experimentally.

In the present study, flash x-ray photography as well as high speed regular photography was used to investigate the fragmentation process during the vapor explosion of single drops of molten metal in water. For drops heated to 700$ sp circ$C, a vapor bubble is formed around the drop after a short induction time of about 80 $ mu$s which grows to a maximum diameter and then collapses. X-ray radiographs, which can "see through" the vapor bubble, show that filaments of metal protruding from the drop surface are produced during the bubble growth. At the time of the bubble collapse, the drop surface is highly convoluted and a second interaction is initiated producing an even larger second bubble in the case of tin drops. A mechanism based on nonuniform superheating and evaporation of the water upon unstable film collapse is proposed to describe the origin of the melt filaments. For liquid drops under isothermal conditions with the water, fragmentation occurs as mass is stripped off the surface due to relative motion of the water. For hot drops at higher flow velocities (i.e., $>$45 m/s) initially a vapor bubble forms but it is quickly displaced downstream. It is proposed that evaporation at the upstream surface of the bubble generates a vapor flow around the drop surface which strips the drop surface. The vapor flow carries the fragments into the bubble region behind the drop. The vapor later condenses leaving a cloud of fragments which are subsequently dispersed by the water. The dynamics of the vapor bubble produced by the explosion of a hot drop under low flow conditions is modeled using a parametric model incorporating the Rayleigh equation for bubble dynamics and an energy equation for the vapor to account for added heat transfer from the distorting drop surface. Estimates of the total heat transferred from the drop show that only 3% of the initial drop thermal energy is converted into work done by the vapor bubble.

This thesis presents a theoretical and experimental study of the stability of cantilevered coaxial cylindrical shells conveying incompressible, viscous fluid inside the inner shell and/or in the annulus between the two shells. Two analytical models are developed with experimental verification. / In the first model, fluid viscous effects are partially taken into consideration. Both shells are generally considered flexible. Shell motions are described by Flugge's shell equations, modified to take into account the steady viscous loads--flow pressurization and skin friction--acting on the shells. These equations are solved by means of the extended Galerkin method, in which the shell equations and the free-end boundary conditions can be satisfied simultaneously. The unsteady viscous forces are approximated by their inviscid counterparts, the formulation of which is based on linearized potential-flow theory with the assumption that the fluid is inviscid. The solution for these forces is obtained with the Fourier-transform technique; in connection with this technique, different so-called out-flow models are examined, concerning the effect of the downstream flow perturbations on the dynamics of the system. / The second analytical model, on the other hand, fully accounts for the viscous effects of the flow. Here, only the inner shell is flexible, while the outer shell is replaced by an identical rigid cylinder. Shell motions are also described by Flugge's modified shell equations, which incorporate the steady viscous loads exerted on the shell. These equations are solved numerically with the finite-difference method. The unsteady viscous forces are evaluated from flow perturbations which are the solution of the linearized, unsteady Navier-Stokes equations subject to the divergence-free constraint on the flow velocity perturbation. A recently developed, time-marching finite-difference method using "artificial compressibility" is applied to solve the Navier-Stokes equations; for the problem under consideration, this method employs the pressure and velocity perturbations as the dependent flow variables on a staggered grid. / In the experimental part of the thesis, tests involving either annular or inner flow are conducted on cantilevered silicone rubber shells concentrically located within rigid plexiglas cylinders. Measurements are made of (i) the critical flow velocity of the system for various lengths of the shell and annular widths, and (ii) the dominant frequencies of oscillation of the shell for certain selected cases. Both divergence- and flutter-type instabilities are observed. / Comparisons between analytical results and test measurements show that the agreement between experiment and the two proposed analytical models is generally good, both qualitatively and quantitatively, in terms of the overall (lowest) critical flow velocities and frequencies of oscillation (first model only) of the tested shells. / Finally, future work is suggested with regard to improving the second model and conducting further calculations.

A class of compliant-joint manipulators is investigated, comprising of an open kinematic chain with revolute joints modeled as linear torsional springs. Singular perturbation techniques are used to separate the slow dynamics from the fast dynamics. The model shows the effect that inertial parameters have on the dynamics of the fast subsystem. / Joint-torque control laws are examined. For large values of drive inertia, it may not be possible to use a fixed torque control law. For sufficiently small drive inertias there always exists a fixed decentralized joint-torque control law that will asymptotically stabilize the fast dynamics. An analysis of high gain joint-torque control laws is presented. / Acceleration feedback is examined. High gain acceleration feedback from collocated drive angles does not destabilize the fast dynamics. However high gain acceleration feedback from noncollocated link angles can destabilize the fast dynamics. Torque control can be used here to stabilize the fast dynamics. Robust control laws for acceleration feedback are discussed.

A modeling technique for simulating the fatigue behaviour of laminated composite materials with or without stress concentrations, called progressive fatigue damage modeling, is established. The model is capable of simulating the residual stiffness, residual strength and fatigue life of composite laminates with arbitrary geometry and stacking sequence under complicated fatigue loading conditions. / The model is an integration of three major components: stress analysis, failure analysis, and material property degradation rules. A three-dimensional, nonlinear, finite element technique is developed for the stress analysis. By using a large number of elements near the edge of the hole and at layer interfaces, the edge effect has been accounted for. Each element is considered to be an orthotropic material under multiaxial state of stress. Based on the three-dimensional state of stress of each element, different failure modes of unidirectional ply under multiaxial states of stress are detected by a set of fatigue failure criteria. An analytical technique, called the generalized residual material property degradation technique, is established to degrade the material properties of failed elements. This analytical technique removes the restriction of the application of failure criteria to limited applied stress ratios. Based on the model, a computer code is developed that simulates cycle-by-cycle behaviour of composite laminates under fatigue loading. / As the input for the model, the material properties (residual stiffness, residual strength and fatigue life) of unidirectional AS4/3501-6 graphite/epoxy material are fully characterized under tension and compression, for fiber and matrix directions, and under in-plane and out-of-plane shear in static and fatigue loading conditions. An extensive experimental program, by using standard experimental techniques, is performed for this purpose. Some of the existing standard testing methods are necessarily modified and improved. To validate the generalized residual material property degradation technique, fatigue behaviour of a 30-degrees off-axis specimen under uniaxial fatigue loading is simulated. The results of an experimental program conducted on 30-degrees off-axis specimens under uniaxial fatigue show a very good correlation with the analytical results. To evaluate the progressive fatigue damage model, fatigue behaviour of pin/bolt-loaded composite laminates is simulated as a very complicated example. The model is validated by conducting an experimental program on pin/bolt-loaded composite laminates and by experimental results from other authors. The comparison between the analytical results and the experiments shows the successful simulation capability of the model.

In this investigation the behavior of pin-loaded composite plates is studied analytically. A progressive damage model is presented which is capable of predicting the three different mechanisms of failure: bearing, shearout, and net tension. The model consists of three major parts: stress analysis, and material property degradation rules. / Based on the model a computer code is developed. The computer code is capable of assessing damage, evaluating residual strength, and predicting ultimate strength of pin-loaded composite plates. Predicted results are compared with available experimental data. Excellent agreement between the predicted and the experimental data was found. / The computer code is used to study geometric parameters that influence joint strength. Such studies are useful in designing mechanical fastened joints using advanced composites.

This thesis presents a mathematical process model for the dynamic analysis of a vertical reheat boiler and the application of this model to the optimal design of a drum water level controller. A numerical finite difference technique is used to formulate this model. / The control system contains two loops, a feedfoward loop using the steam flow and drum pressure signals as the input and a feedback loop using the deviation of the measured drum water level from its set point as an input. The feedback loop is an incremental PID controller with an adjustable proportional gain. The feedfoward loop is designed to directly actuate the control devices before the "swell" and "shrinkage" in the boiler water level occur. The feedforward controller output signal is summed along with the output of the PID controller to establish the set point for the control actuator. This scheme is effective because steam flow changes are immediately fed forward to change the final feedwater set point for the control actuator. In this way, feedwater flow tracks steam flow and any disturbances in the feedwater system will be arrested quickly. / It is shown that an incremental PID controller plus adapt feedfoward compensator can be successfully employed for the control of water level in such a plant.

No description available.

Potential difficulties involved in Computational Fluid Dynamics (CFD) approaches to the investigation of primary-flow domains linked by multiple orifices are identified. Practices that involve detailed simulation of the complete flow field, including flow through the orifices, are limited by the need for extensive grid refinement near each orifice, and by slow convergence rates. Other approaches, such as modeling of the flow in each domain separately, and linking these flows by one-dimensional flow networks, reduce the accuracy of the solution. In this thesis, an alternative treatment is developed in which finite-volume modeling (FVM) inside the slot is replaced by specified flow conditions at the slot boundaries. These conditions are updated once per iteration by a correlation function, based on flow conditions local to the slot, as generated by the primary-domain FVM solution. The correlation function is derived empirically, using data from detailed numerical simulations. Application of this approach to laminar flow in two parallel channels linked by a small slot has produced encouraging results.

Over the past decade, theoretical investigations have revealed the possible existence of two distinct mechanisms, a fluid-damping controlled one requiring only a single degree-of-freedom system and a fluid-stiffness controlled one requiring two or more degree-of-freedom system, instrumental in causing fluidelastic instability of cylinder arrays subjected to fluid cross-flow. As yet, the existence of these mechanisms has not been verified experimentally, and some researchers tend to neglect one or the other of these mechanisms in their theoretical studies. / In this thesis, with the objective of obtaining further insight into the nature of fluidelastic instability mechanisms, experimental and theoretical studies have been performed on a rotated square array with $P/d$ = 2.12. Previous theoretical and experimental studies on this array have established the fact that a single flexible cylinder, in an otherwise rigid array, is fluidelastically stable. However, multiple flexible cylinder dynamic (vibration) experiments undertaken in this study show that fluidelastic instability develops when the array incorporates three of more flexible cylinders. This result verifies the duality of the instability mechanisms and suggests that the cylinder motion in the present array is dominated by the fluid-stiffness controlled mechanism, rather than the fluid-damping controlled mechanism. / Involved dynamic (vibration) experiments have been undertaken to elucidate the effect of various parameters such as, number of cylinders, cylinder position, cylinder mass, frequency detuning and fluidelastic coupling on the instability threshold of this array, in which the fluid-stiffness controlled mechanism prevails. It has been determined that varying mechanical damping has a small effect on the critical velocity, whereas, varying cylinder mass generates, relatively, large changes in the critical velocity. A "Connors type" instability equation, or versions of it, are shown not to be applicable in this array, mainly due to the strong dependence of the mass exponent on the actual value of the non-dimensional mass. / Frequency detuning of adjacent cylinders is also shown to have a significant effect on the critical velocity. Further dynamic (vibration) experiments revealed the co-existence of dynamic and static instabilities within close proximity to each other. It was possible to switch from one type of instability to the other, by varying one, or more, of the mechanical properties of the flexible cylinders. / Next, the time averaged fluid forces acting on static cylinders were measured as a function of monitored, and surrounding, cylinder displacements at different Reynold numbers, to attain a physical understanding of the flow pattern in the array. The results complemented and verified the various dynamic and static instability findings of the vibration (dynamic) experiments. / Finally, the fluid forces were incorporated in a quasi-steady, multiple degree-of-freedom model for comparison with experimental results.