Pannasit Borwornpiyawat.. Development of the in-house CFD solver for steady laminar incompressible flow on structured and unstructured meshes in a skewed cavity. Master's Degree(Mechanical and Automotive Engineering (International Program)). King Mongkut's University of Technology North Bangkok. Central Library. : King Mongkut's University of Technology North Bangkok, 2016.
Development of the in-house CFD solver for steady laminar incompressible flow on structured and unstructured meshes in a skewed cavity
Abstract:
In order to improve the performance of the in-house CFD solver for the steady laminar incompressible flow in terms of robustness and accuracy especially when the grid is non-orthogonal, the Coordinate Transformation Method (CTM) and Surface Decomposition Method with Improve Deferred Correction Scheme (SDM-IDC) are implemented into the in-house solver for 2D structured and unstructured meshes. It was revealed that the CTM method can perform better than the SDM method when applied to solve the Navier-Stokes equations (Borwornpiyawat, et al., 2016). However when applied to 3D problem, only the SDM method is applicable while the original CTM method cannot be used. In attempt to achieve higher accuracy in 3D problem, a new method called Revised Surface Decomposition Method (rSDM) is developed by combining the techniques from both methods together. The performance of the in-house CFD solver is then evaluated by solving the flow in a skewed cavity, in which case the skew angle ranges from 0 to 89 degrees at the Reynolds number of 100 and 1,000. The results from ANSYS Fluent are also presented for comparison. From this study, it was revealed that the CTM method is very robust since it can converge even at the angle of 89 degrees whereas the SDM-IDC method and ANSYS Fluent fail to converge when the angle is greater than 45 degrees. At the Reynolds number of 1,000, the CTM method is more accurate than the SDM-IDC method but becomes slightly less accurate when the Reynolds number is equal to 100. For 3D problems, the accuracy of the rSDM method is equivalent to the original SDM method but consumes more computational time. For this reason, the rSDM method is not included in the final version of the in-house solver
King Mongkut's University of Technology North Bangkok. Central Library