This exercise comprises of two sections A and B, where in the first section an analysis in creating initial vertices will be carried out along with the creation of edges and faces, and the setting of boundary types. The program used for this phase of the investigation will be Gambit. These will then be generated into a mesh using Fluent. Moreover, section B will focus on the sensitivity to the computational mesh, sensitivity to the residuals and the use of higher order numerical discretization schemes. The structure of this report analyses the discretization of meshes, showing difference in results subject to variable changes. The general findings show that the different contours and iterations produced become increasingly accurate as the grid value increase or the order of discretization increases. 2. Altering the Residuals

The following graphs show comparisons in residuals for different number of iterations, for a first order discretization 2.1 Iteration for a triangular mesh

{draw:frame} {draw:frame}

Figure 2.1: Original mesh Figure 2.2:Mesh with reduced residuals 2.2 Iteration for a quadilateral mesh

{draw:frame} {draw:frame}

Figure 2.3: Original mesh Figure 2.4: Mesh with reduced residuals From the graphs above one can observe an increase in iteration when the residuals are reduced. However, the general shape of the curves remain exactly the same for both the triangular and quadrilateral mesh. From these findings it can be concluded that as the convergent rate decreases, the number of iterations becomes more important. When the convergence was set at a value less than 10e-04 the results were more accurate. *3. Comparison For* First Order Discretization

3.1 Velocity magnitude

{draw:frame} {draw:frame}

Figure 3.1: Velocity variation for triangular mesh Figure 3.2: Velocity Variation for quadrilateral mesh These graphs show that the diffusivity isn’t as well accounted for the quadrilateral mesh as it is for the triangular mesh. This is evident from the greater amount of uniformity and higher magnitude in the quadrilateral mesh than the triangular one. 3.2 Static Temperature

{draw:frame} {draw:frame}

Figure 3.3 Static temperature for triangularmesh Figure 3.4 Static temperature for quadrilateral mesh Likewise to the previous section, the quadrilateral mesh displays more uniform, refined lines than the triangular one, however in contrast the temperature scale is within the same limits. The triangular mesh shows a more gradual spread in temperature than the quadrilateral one. The limits can clearly be obtained here from using a mesh with fewer edges since there is less precise detail in the triangular mesh, i.e. it is more difficult to make out the colours separately. 3 Velocity Vectors The 2 diagrams below show a similarity in the velocity vectors for both the meshes; however where the quadrilateral mesh presents a more definite, uniform profile, the triangular mesh shows a less organized and somewhat scattered group of lines. Again this only further proves the greater precision in having a larger number of edges. {draw:frame} {draw:frame}

Figure 3.5 Velocity vector for a triangular mesh Figure 3.6: Velocity magnitude for a quadrilateral mesh 3.4 Static Pressure

{draw:frame} {draw:frame}

Figure 3.7 Static pressure contours for triangular mesh Figure 3.8 Static pressure contours for quadrilateral mesh For the static pressure variable, the quadrilateral mesh appears to have a greater sensitivity than the triangular mesh, displaying an area of high pressure on small inlet pipe and a greater contour profile on the area of high pressure. However what’s interesting to note is the area of low pressure detected at the top curve of the triangular mesh which is completely neglected in the quadrilateral mesh. Furthermore at the points of lowest pressure for both diagrams, there...