Unstructured Grid Generation Technique And Numerical Simulation Of Flow Field

The purpose of this dissertation is to study the technique of unstructured grid generation, numerical simulation of flow field on unstructured grid and adaptive grid generation.Firstly, an unstructured grid generation process based on Delaunay triangulation method, is presented, which is a combination of Kennon's algorithm and automatic point generation technique.Secondly, a new strategy for unstructured grid generation is presented, which is named new Delaunay triangluation method because the element generated by the new method satisfy the property of Delaunay triangle. The new method overcomes successfully all difficulties of traditional Delaunay triangulation algorithm, such as the loss of edge on boundaries and the generation of unwanted triangles. The new algorithm can generate high quality triangles, which will not penetrate through boundaries.Thirdly, three different kinds of software, which are capable of handling arbitrary domain with minimum user interventions, are achieved successfully by means of combining separately the new method with automatic point generation technique, iterative point insertion and Advancing Front Method. In the process of program, reasonable data structure and fast search algorithm are employed to improve the efficiency of computation and save greatly the computer time. Application of these mesh generators to many problems shows them to be the most powerful.Finally, Jameson finite volume scheme and four-stage Runge-Kutta time-stepping method are applied to numerical simulations of flow filed. Local time step and residual smoothing technique are introduced to accelerate the convergence and reduce the computer time effectively. The derivative of density is taken as criterion to detect high gradient regions of the flow field where shocks occur, we can generate high quality adaptive grids. The numerical experiments are made with methods mentioned above, the results show complex inviscid flow field can be simulated successfully by use of adaptive grids...