Product help with community knowledge

Page statistics

1456 views1 edit(s)9719 characters(s) Page last modified 09:34, 18 Dec 2012 by contentconnector

CFD Simulation Considerations

In this topic, we discuss the ideal flow region or wind tunnel size, the implications of 2D vs. 3D simulations, and the effects of mesh resolution on the results.

Wind Tunnel Size

Falcon creates a default wind tunnel based on the dimensions of the model. You should ensure, however, the wind tunnel is large enough to not artificially affect the results, especially if you are simulating an object in free stream flow.

For best results, there should be a gap between the wind tunnel and the model that is two to three times the dimension of the model in that respective dimension:

The wind tunnel should also be large enough in the flow-wise direction. Ideally, the wind tunnel extends at least 2 model lengths upstream and at least 4 model lengths downstream. This distance ensures that the boundary conditions do not artificially affect the flow approaching the model, and it allows the wake downstream of the model to form correctly:

A wind tunnel that is too small confines the flow around the model. The flow accelerates between the walls and the model, which affects the flow results. Remember that the walls bound the simulation, and no flow can pass through them. Here is an example of a wind tunnel that is too close to the model:

As the flow accelerates between the model and the walls, the pressure drops and the flow accelerates. If the physical object is not near any such boundaries, this effect is artificial, and the simulation results are incorrect.

Each Falcon product uses a slightly different way to adjust the wind tunnel size. Please see the Sizing topic for your particular Falcon version.

2D and 3D Simulations

There are two fundamental simulation modes you can select: 2D and 3D. You should choose the mode that works best for your simulation goals and accuracy objectives.

2D Simulation runs the fastest by computing results on a two-dimensional slice through the model. Use it to quickly understand the aerodynamic behavior through a specific cross-section.

Flow in 2D mode, however, cannot move out of the plane to pass around an obstruction. It must "bunch" up to one side or the other because it is constrained to the plane. In this example, as the air hits the front of the model, it turns upward and downward to flow past it. In reality, we know that some of the air would also flow past the sides of the model:

In most cases, this is not a physical representation of the flow, so the results might not be as accurate as in a 3D simulation. Does this mean that 2D results have no value? Not at all. Because 2D simulations solve quicker than 3D, they provide a valuable concept view of the flow. The 2D mode is a powerful way to conduct multiple "what if" scenarios early in the design process. As the design evolves, however, it is wise to verify the flow behavior with a 3D simulation.

3D Simulation is a more rigorous way to simulate flow aerodynamics. It is useful for visualizing the interactions between the flow and the model. Results computed in 3D are more complete than in 2D because they encompass the entire object and the air throughout the wind tunnel. When moving air encounters an obstruction, it can pass over, to the side, or under, as it would physically. In the following, some of the air flows past the sides, so not as much is forced over and under the model:

Summary:

• Results computed in 2D mode occur only in the selected plane. Use 2D for conceptual level "what if" simulations. Note that 2D does not produce a complete view of the flow.
• Results computed in 3D mode are more complete. 3D encompasses the entire object and the air throughout the wind tunnel. 3D allows the air to move in any direction that it would physically. Use 3D to evaluate the true aerodynamics of your model.

Mesh Resolution

The calculation model is transparently broken into small pieces called a mesh. This is necessary for the simulation engine, and Falcon computes the mesh distribution automatically.

In some cases, however, it may be necessary to refine the mesh to better capture complex flow distributions around complex geometric features. This is an example of a mesh that is too coarse to produce good results:

When the mesh is refined, the result are much clearer: