Available at: http://digitalcommons.calpoly.edu/theses/337
Date of Award
MS in Aerospace Engineering
Dr. Robert A. McDonald
The utility of computational fluid dynamics (CFD) for solving problems of engineering interest has experienced rapid growth due to the improvements in both memory capacity and processing speed of computers. While the capability now exists for the solution of the Navier-Stokes equations about complex and complete aircraft configurations, the bottleneck within the process is the time consuming task of properly generating a mesh that can accurately solve the governing partial differential equations (PDEs). This thesis explored two numerical techniques that attempt to circumvent the difficulty associated with the meshing process by solving a simplified form of the continuity equation within a meshless framework. The continuity equation reduces to the full potential equation by assuming irrotational flow. It is a nonlinear PDE that can describe flows for a wide spectrum of Mach numbers that do not exhibit discontinuities. It may not be an adequate model for the detailed analysis of a complex flowfield since viscous effects are not captured by this equation, but it is an appealing alternative for the aircraft designer because it can provide a quick and simple to implement estimate of the aerodynamic characteristics during the conceptual design phase.
The two meshless methods explored in this thesis are the Dual Reciprocity Method (DRM) and the Generalized Finite Difference Method (GFD). The Dual Reciprocity Method was shown to have the capability to solve for the two-dimensional subcritical compressible flow over a Circular Cylinder and the non-lifting flow for a NACA 0012 airfoil. Unfortunately these solutions were obtained with the requirement of a priori knowledge of the solution to tune a parameter necessary for proper convergence of the algorithm. Due to the shortcomings of applying the Dual Reciprocity Method, the Generalized Finite Difference Method was also investigated. The GFD method solves a PDE in differential form and can be thought of as a meshless form of a standard finite difference scheme. This method proved to be an accurate and general technique for solving the previously mentioned cases along with the lifting flow about a NACA 0012 airfoil. It was also demonstrated that the GFD method could be formulated to discretize the full potential equation with second order accuracy. Both solution methods offer their own set of unique advantages and challenges, but it was determined that the GFD Method possessed the flexibility necessary for a meshless technique to become a viable aerodynamic design tool.