You are on page 1of 5

Waqas Saleem

090341939

Introduction
The purpose of this exercise was to use computational approach towards solving a T-junction fluid flow problem. CFD software ICEM and Fluent were used to examine the fluid flow in the junction mentioned in Figure 1. ICEM was used to create the basic geometry and meshes (figures 2, 3 &4). These were then exported to FLUENT for running simulations, which were used to draw conclusions about the behavior of the flow. Fluid properties were set for water-liquid flowing at a constant value of 0.05cm/s.

outlet2
1 cm

Wall2 Wall1
1 cm

Wall3 Wall4
20 cm

10 cm

inlet
0.05 cm/s

outlet1

Wall5

Figure 1

Assessing the accuracy of the solution


The accuracy of a solution obtained from the use of CFD software can be assessed by the solution in subject with a present solution of the similar sort. Literature can be read for similar problems in order to have an insight on ways for obtaining higher accuracy by improving the modeling. Moreover, experience in CFD simulations is very important for obtaining comparatively more accurate solutions than an amateur user.

Effect of increasing mesh density


Mesh density affects the truncation error in the simulation. By increasing the mesh density the solution becomes more accurate. However, it is important to choose an appropriate value for the number of nodes in the mesh by performing the mesh sensitivit y analysis. An insufficiently dense mesh would give inaccurate solutions while an excessively denser mesh would be a waste of time, as it would take far too long to simulate the problem. By reducing the cell sizes in ICEM the mesh density increased and the results became more accurate. For this exercise meshes with 21, 31 and 41(Figures 2, 3 & 4) nodes were created in ICEM and then the FLUENT results confirmed that the accuracy increased with increasing number of nodes. Due to the page limit constraints, this report includes FLUENT results obtained from mesh (created in ICEM) with 31 nodes only.

Figure 2

Figure 3

DEN331: Computer Aided Engineering for Solids and Fluids

Waqas Saleem

090341939

Figure 4

Influence of choice of scheme on the accuracy of the solution


It was observed that the solution became more accurate as the order of scheme was increased. 1st order is the simplest numerical scheme. It resulted in very stable calculations, however the gradients in the flow field appeared to be smeared out. The 2nd order scheme was more accurate than the 1st order, but in regions with strong gradients it resulted in face values that were outside the range of cell values. To correct the error, limiters were applied to the predicted values. From the observations made during the simulations, it was concluded that it is recommended to start with the first order and then switch over to the second order after a reasonable number of iterations in order to achieve a balance between stability and accuracy.

Unstructured and structured grids


Structured grids are comprised of families of intersecting lines, one in each space dimension. Each mesh point is located at the intersection of a line from each family. On the other hand, unstructured grids have arbitrary distributions of mesh points. The points in unstructured grids are connected by triangles, quadrilaterals or polygons. The selection of grid type varies by the problem geometry and fluid type. If the geometry is complex or the fluid behavior is uneven, then it is advisable to use unstructured grids in order to achieve a more accurate solution. For this particular exercise, structured grids were used while creating the mesh as the problem geometry is fairly simple and the fluid in subject (water) is a Newtonian fluid.

Visualization of the flow field in light of static pressure contours and the velocity vector fields

Figure 5 Pressure Contour (0.05 cm/s)

DEN331: Computer Aided Engineering for Solids and Fluids

Waqas Saleem

090341939

Figure 6 Velocity Vector (0.05 cm/s)

Shear stresses on wall 3 & 4

Figure 7 Stresses Wall 3 (0.05 cm/s)

Figure 8 Stresses Wall 4 (0.05 cm/s)

DEN331: Computer Aided Engineering for Solids and Fluids

Waqas Saleem

090341939

Changing the flow split ratio between outlets


Varying the flow rate weighting in the boundary conditions changed the flow split ratio between both outlets. This feature is very useful when simulating complex problems as it allows the user to achieve desired mass flow rate between different outlets. For example, if there is a suction pump at one end which increases the flow rate in a particular outlet. This would reduce the flow rate coming out from the other outlet(s). Alternating these values would simulate this scenario.

Comparison of the current flow with a higher flow rate


By increasing the velocity of the flow from 0.05cm/s to 1cm/s the flow in outlet 1 was increased while the flow from outlet 2 decreased significantly. It was also observed that as the velocity reached the junction, it got split and the magnitude of the velocity vectors reduced. As for the wall stresses, it was observed that by increasing the flow velocity, the stresses were increased on wall 4 and reduced on wall 3 as a consequence of increased flow rate (and velocity) through outlet 1. Figures 9, 10, 11 and 12 show the pressure contours, velocity vectors and wall sheer stresses as a consequence of the increased flow velocity.

Figure 9 Pressure Contour (1cm/s)

Figure 10 Velocity Vector (1cm/s)

DEN331: Computer Aided Engineering for Solids and Fluids

Waqas Saleem

090341939

Figure 11 (1cm/s)

Figure 12 (1cm/s)

Conclusion
Upon the completion of this exercise it was concluded that the accuracy of a simulation can be increased by choosing an appropriate mesh size, increasing the choice of scheme and having a suitable choice of grid type which favors the fluid type and geometry of the object. Moreover, it was observed that the fluid velocity had a significant effect on the behavior of the fluid flow.

References
1. Oleg Zikanov, (2010). Essential Computational Fluid Dynamics. 1 Edition. Wiley 2. Andre Bakker. (2002). Introduction to CFD. Available at: http://www.bakker.org/dartmouth06/engs150/01intro.pdf [Accessed on 08 January 2012]Mechanical 3. Engineering. (2011). Computational Fluid Dynamics: Principles and Applications. Available at:http://www.mechanicalengineeringblog.com/tag/computational-fluid-dynamics/[Accessed on 09 January 2012] 4. CFD Online. (2011). Fluent FAQ: Structured Grids. Available at: http://www.cfdonline.com/Wiki/Fluent_FAQ#Q:_How_do_I_determine_what_kind_of_grid_to_generate.3F_.28I_would_li ke_to_known_selection_criteria_between_structured_and_unstructured_mesh..29 [Accessed on 10 January 2012] 5. Yong Sheng Khoo. (2009). Cornel University: FLUENT: Steady Flow Past a Cylinder Step 7. Available at: https://confluence.cornell.edu/display/SIMULATION/FLUENT+-+Steady+Flow+Past+a+Cylinder++Step+7 [ Accessed on 10 January 2012]

DEN331: Computer Aided Engineering for Solids and Fluids

You might also like