Just a few years ago, the use of computational fluid dynamics (CFD) in most aerospace companies was restricted to pure research or troubleshooting problems with existing designs. But in the past few years, newly available CFD tools are fully embedded in the mainstream mechanical design environment and hence much easier, faster and less expensive to use. The trend toward using CFD early stages in the design process has resulted in a large number of new users of this powerful analysis technique. Bell Helicopter technical specialists have developed a set of CFD best practices with the goal of guiding users in following company design guidelines and optimizing the analytical processes. The tools are based on guidelines developed for use by engineers working on propulsion systems. The guidelines are based on the application of Mentor Graphics Mechanical Analysis Division’s FloEFD suite of CFD software used by Bell Helicopter for internal flow simulations.
These best practices empower design and analysis engineers to analyze concepts during the first gates of the design process, so that problems can be corrected early and performance can be optimized at the lowest possible cost. These CFD tools utilize native 3D data and provide automatic gridding of the flow space, minimizing the need for engineers to expend significant effort on the numerics of CFD.
Verifying Solid Model Quality
The utilization of native 3D data places a premium on the quality of the solid model. For an internal flow model with minimum mesh requirements, the solids must form a sealed internal space with no leak paths outside the internal flowfield. The flow boundary conditions should be applied to “lids” that form the inlet faces and outlet faces of the computational domain. Minute details of the geometry should be eliminated wherever possible to keep the CFD model size to a minimum. For example, latches, fasteners, and brackets on the inner surface of cowling should be eliminated unless they are critical to the flow field in order to reduce the CFD solution time. After the geometry is imported, it should be checked for problems using the “check geometry” feature in the CFD software. Any invalid contacts listed must be resolved prior to analysis.
Typically, invalid contacts occur when two parts in the assembly share an edge. Also, for internal flow problems, the flow volume value at the end of the check must be a positive number. A zero flow volume indicates a leak path out of the internal flow field that should be resolved. Thin-walled solids can result in irregular cells if the mesh density is too low. Irregular cells computationally represent a hole in a thin solid. If there are numerous irregular cells in a mesh, then a solid wall will no longer provide a physical barrier in the simulation. Irregular cells can be remedied by increasing the mesh density in the area with the irregular cells. This can be accomplished by increasing the global mesh level, shrinking the minimum wall thickness in the mesh settings, or applying a local mesh in the affected zone to impose a higher level of cell refinement.
Transient simulation is particularly challenging from a computational standpoint because the flow field must be solved for a number of time steps. These guidelines have been developed by Bell technical specialists to overcome the challenges of transient analysis:
- Minimize geometry detail; exclude volumes without flow boundary conditions; keep cell count at a minimum.
- Ensure simulation initial conditions are accurate – these are critical for transient simulations.
- Double-check boundary and free - stream conditions – transient simulations potentially may take days to complete, so ensure everything is absolutely perfect before launching a simulation.
- Establish an equation goal to track and record physical time at during every iteration.
- Establish a solution file output period consistent with the file size, flow gradients, and physical time duration of analysis.
- Use the default physical timestep size initially to accelerate the solution ramp up to larger time steps as the simulation progresses if initial high flow gradients dissipate.
Checking For Sensitivities
The mesh should be refined such that the solution does not depend on the mesh size. To ensure that there is no mesh spacing dependence, conduct a mesh sensitivity study at least once during each simulation project. This study should be performed by increasing or decreasing the mesh size and checking to see if the solution changes significantly. Another option is to activate solution-adaptive meshing and allow the number of cells to automatically increase in areas of high flow gradient, and again ensure the flow solution remains unchanged for the purposes of the analysis. External flow simulations are used to analyze areas such as inlet recovery and exhaust gas impingement. The computational domain size is critical for this type of simulation. It must be large enough that computational boundaries do not influence the flowfield in the areas of interest. A computational domain sensitivity study should be carried out with a larger computational domain to ensure that the results of interest are not influenced by the far-field boundaries.
In CFD analysis of propulsion systems, it is important to identify and resolve any problematic flow separation. Some ways flow separation can be identified include:
- Look for sharp direction changes or swirls in cut plots showing velocity contours and vectors.
- Release a high concentration of streamlines upstream of an area with diverging flow area, adverse pressure gradient, or change of flow direction, and look for swirling or sudden changes of direction in the streamlines.
- Create an isosurface at zero velocity in the primary flow direction to identify and visualize the flow separation zone in 3D. Typical methods of resolving flow separation include eliminating surface irregularities, reducing the rate of area expansion, or increasing the internal radius of a lip or bend.
Here’s an example of how these methods have been used to solve real-world design challenges. Some aircraft include an inerting system that introduces nitrogen gas into the airspace of the fuel tank to make it less likely to ignite any fuel vapors there should an ignition source be present. The question that needs to be answered by analysis is: how long does a pilot have to wait to take off without compromising the safety of the aircraft; how long until the fuel tank is adequately inerted? The final answer is provided by testing, but each physical test can be very expensive to perform. The problem was handed off by a product team lead to an engineering analyst who had never done this type of analysis, but made good use of the best practice guidelines. She used the detailed work procedures in the best practices to check the geometry and make modifications where needed. She obtained the initial conditions from the integrated product team and added them as boundary conditions on the CAD solid model. She created the mesh and checked it for adequate resolution.
She then solved the model and generated output including a chart that shows oxygen concentration in the tank over time, then confirmed with further analysis that the solution was not mesh-dependent. The CFD results were compared and validated to preliminary bench test data. With confidence in the results and analytical approach, the geometry can be optimized and multiple design concepts can be evaluated prior to any performance tests. It is far less expensive and less time consuming performing analytical iterations to determine the best design than planning and conducting tests of several design concepts, each of which must be specially fabricated solely for test purposes. The analysis in the early phases of the design process allowed for design optimization, provided confidence to engineers and customers that the design is robust.