Development of asymmetric 3d shock prediction algorithm for high speed flows

Shock waves are formed as a result of the combination of disturbances around vehicles moving at supersonic speeds. These waves are generally characterized by sudden and discontinuous changes in pressure, temperature, and density in the medium. CFD applications are frequently used in solving complex problems with strong gradients, such as shock phenomena. Nowadays, one of the main focuses of CFD is to solve complex problems in a minimum amount of time while maintaining the fidelity of the analyses. For supersonic and hypersonic flow problems, an effective way is to generate a suitable computational mesh that is refined and aligned along the strong gradient regions, like the shock waves. In addition, correct prediction of the shock shape is required to use the fast flow prediction methods, which are frequently used in design problems. Therefore, different approaches have been developed in the literature to determine the shock shapes. Some of the common approaches can be presented as analytical equation-based methods that can only be used for simple geometries, methods based on the inverse solution of simplified conservation equations, and mesh adaptation methods based on predetermined solutions. In this study, a faster model than previously defined methods in the literature was developed to accurately predict stationary shock waves around high-speed aircraft without the use of expensive CFD simulation or simulation examination costs. First of all, the validation studies of open-source SU2 v7.1.1 software, which is used in CFD applications, were performed on spherically blunt-nosed geometries at different angles of attack (0, 5, 10.25). Mesh-independent solutions were determined by using Spalart-Allmaras and SST SST k-w turbulence models. As a result of the calculations, it was observed that the CFD results agree well with the experimental data. Two and three-dimensional shock shape prediction algorithms were examined in detail within the scope of the thesis. The 2D shock shape prediction method mainly consists of three parts, which are: detached shock prediction with Billig experimental hyperbola curves; attached shock prediction with analytical shock-expansion equations; and a sonic line correction process using the CFD correction factor. The 3D shock prediction method is basically designed as an application of the 2D methodology on surface streamlines. In order to calculate the surface streamlines, the geometry is expressed with a surface mesh. After that, the free flow velocity is reflected on the surface, and the fluid kinematic equations are solved over the surface mesh. The calculated shock points are expressed as 3D shock geometry, and the shock shape prediction process is complete. Validation studies were performed for 2D and 3D shock shape prediction algorithms. In validation studies, 2D and 3D algorithms were compared with HAD and experimental results using ogive and spherically blunt-nosed geometries. CFD analyses were performed at Mach 5 for ogive and Mach 3, 5, 8 for blunt-nosed geometries. Free stream conditions for analyses were examined at angles of attack of 0, 5, 10 degrees. The shock shape determined from the CFD analysis and the shock geometry obtained as a result of the algorithm were compared using their the vertical distance with respect to the geometry exposed to the flow. As a result of the 2D and 3D investigations, almost no difference was observed in the ogive geometry, while it was observed that the percentage differences increased as the angles of attack increased in the conical geometry comparisons. While the maximum difference was calculated as approximately %1 in 2D validation studies, the difference can increase up to %4 in 3D studies. In addition to the comparison with the CFD calculations, when the 3D shock shape prediction method was compared with the experimental shock shape, it was observed that the errors increase with the angle of attack as determined from the CFD comparisons. The maximum error between the experimental data and the model was calculated as %3.