A three-dimensional unstructured incompressible RANS code has been developed using artificial compressibility and Spalart-Allmaras eddy viscosity model. A node-based finite volume method is used in which all flow variables are defined at the vertices of tetrahedrons in an unstructured grid. The inviscid fluxes are computed by using the Roe's flux difference splitting method, and higher order accuracy is attained by data reconstruction based on Taylor series expansion. Gauss theorem is used to formulate necessary gradients. For time integration, an implicit scheme based on linearized Euler backward method is used. A tetrahedral unstructured grid generation code has been also developed and applied to the tip clearance flow in a highly staggered cascade. Surface grids are first generated in the flow passage and blade tip by using several triangulation methods including Delaunay triangulation, advancing front method and advancing layer method. Then the whole computational domain including tip gap region is filled with prisms using the surface grids. Each prism is divided into three tetrahedrons. To accomplish this division in a consistent manner, connectivity pattern is assigned to each triangle in the surface grids. A new algorithm is devised to assign the connectivity pattern without reference to the particular method of triangulation. This technique offers great flexibility in surface grid generation. The code has been validated by comparisons with available computational and experimental results for several test cases: invisicd flow around NACA section, laminar and turbulent flow over a flat plate, turbulent flow through double-circular arc cascade and laminar flow through a square duct with 90° bend. For the laminar flat plate case, the velocity profile and skin friction coefficient are in excellent agreement with Blasius solution. For the turbulent flat plate case, velocity profiles are in full agreement with the law of the wall up to Reynolds number of 1.0E8, however, the skin friction coefficient is under-predicted by about 10% in comparison with empirical formula. Blade loading for the two-dimensional circular arc cascade is also compared with experiments. The results obtained with the experimental inflow angle (51.5° ) show some discrepancies at the trailing edge and severely under-predict the suction peak at the leading edge. These discrepancies are completely remedied if the inflow angle is increased to 53.5° . The code is also capable of predicting the secondary flow in the square duct with 90° bend, and the velocity profiles are in good agreement with measurements and published Navier-Stokes computations. Finally the code is applied to a linear cascade that has GE rotor B section with tip clearance and a high stagger angle of 56.9° . The overall structure of the tip clearance flow is well predicted. Loss of loading due to tip leakage flow and reloading due to tip leakage vortex are presented. On the end wall, separation line of the tip leakage vortex and reattachment line of passage vortex are identified. The location of the tip leakage vortex in the passage agrees very well with oil flow visualization. Separation bubble on the blade tip is also predicted. Mean streamwise velocity contours and cross sectional velocity vectors are compared with experimental results in the near wake, and good agreements are observed. It is concluded that Spalart-Allmaras turbulence model is adequate for this type of flow field except at locations where the tip leakage vortex of one blade interacts with the wake of a following blade. This situation may prevail for blades with longer span and/or in the far wake. Prediction of such an interaction presents a challenge to RANS computations. The effects of blade span on the flow structure have been also investigated. Two cascades with blades of aspect ratios of 0.5 and 1.0 are considered. By comparing pressure distributions on the blade, it is shown that the aspect ratio has strong effects on loading distribution on the blade although the tip gap height is very small (0.016 chord). Grid convergence study has been carried out with three different grids for pressure distributions and limiting streamlines on the end wall. / Ph. D.

Population-wide synchronized rhythmic bursts of electrical activity are present in a variety of neural circuits. The proposed general mechanisms for rhythmogenesis are often attributed to intrinsic and synaptic properties. For example, the recurrent excitation through excitatory synaptic connections determines burst initiation, and the slower kinetics of ionic currents or synaptic depression results in burst termination. In such theories, a slow recovery process is essential for the slow dynamics associated with bursting. This thesis presents a new hypothesis that depends on the connectivity pattern among neurons rather than a slow kinetic process to achieve the network-wide bursting. The thesis begins with an introduction of bursts of electrical activity in a purely excitatory neural network and existing theories explaining this phenomenon. It then covers the small-world approach, which is applied to modify the network structure in the simulation, and the Morris-Lecar (ML) neuron model, which is used as the component cells in the network. Simulation results of the dependence of bursting activity on network connectivity, as well as the inherent network properties explaining this dependence are described. This work shows that the network-wide bursting activity emerges in the small-world network regime but not in the regular or random networks, and this small-world bursting primarily results from the uniform random distribution of long-range connections in the network, as well as the unique dynamics in the ML model. Both attributes foster progressive synchronization in firing activity throughout the network during a burst, and this synchronization may terminate a burst in the absence of an obvious slow recovery process. The thesis concludes with possible future work.

The disulfide bond in a protein is a single covalent bond formed from the oxidation of two cysteines. It plays an important role in the folding and structure stability, and may regulate protein functions. The connectivity prediction problem is difficult because the number of possible patterns grows rapidly with respect to the number of cysteines. We discover some rules to discriminate the patterns with high accuracy in many methods. We implement multiple SVM methods, and utilize the BKS to fuse these classifiers. We apply the hybrid method to SP39 dataset with 4-fold cross-validation for the comparison with the previous works. We raise the accuracy to 71.5%, which improves significantly that of the best previous work, 65.9%.