Incompressible 2D flow through an elbow with OpenFOAM

A simple 2D incompressible flow through an elbow was analyzed using OpenFOAM; this post will show the results for two meshes, one with triangularcells and the other with quad cells; a final analysis used a twice refined mesh.

Case setup:
- velocity inlets: the left (1 m/s) and lower (3 m/s) openings;
- pressure outlet: the upper opening, 0 Pa;
- no slip walls: all other edges;
- viscosity: 0.01 m2/s;
- time 35 s (determined after a first run with a longer time).


This case is given in the tutorial files in the OpenFOAM installation;  the given mesh is a triangular Fluent mesh; as it will be seen, the given mesh is not the best for this case and another mesh is used for more reliable results.

The meshes:

Triangular cells

Quad cells

Quad cells - refined

The velocity results for the converged solution:

Velocity for tri-mesh

Velocity for quad-mesh

Velocity for quad-mesh-refined

The pressure results for the converged solution:

Pressure for tri-mesh

Pressure for quad-mesh

Pressure for quad-mesh-refined

Summary and discussion:
This simple case shows the importance of the mesh used in a CFD analysis; as it can be seen from the results, the triangular cell mesh introduce a lot of numerical diffusion in the solution.
Much better results are obtained with the quad mesh because the mesh is aligned with the flow. The velocity plot for the third case shows clearly two recirculations that were missed with the first mesh.

What to take from this post:
- the tutorials are created to show us how the program works and not to teach us how good results are obtained;
- always check the results with different meshes, physical models, etc. and against experiments if available;
- a good practice is to refine the mesh to see if the results are grid independent, like I showed here.

If you have any questions regarding this analysis don't hesitate to leave a comment below.