'Aerodynamics analysis of an F-117 bomber' simulation project by emadarali


#1

I created a new simulation project called 'Aerodynamics analysis of an F-117 bomber':

An aerodynamics study of the unique F-117 design in SimScale.


More of my public projects can be found here.


#2

Introduction

The F-117 stealth bomber is one of the most recognizable and unique aircraft designs of the 20th century. Its distinctive shape is a direct result of the prioritization of the reduction of Radar Cross Section (RCS) over aerodynamic performance during its design. The sharply angled components of its fuselage serve to reflect the incoming radar waves away from the aircraft and the transmitter aircraft. Naturally this causes significant aerodynamic inefficiency and the aircraft is aerodynamically unstable in all three axes, necessitating complicated Fly-By-Wire systems to achieve stable flight.

As a Mechanical Engineer who had a focus on Aerodynamics, I thought this was the perfect project for my first serious SimScale project.

Pre-processing

I chose the GrabCad member Paulo Vitor de Souza Costa’s amazingly detailed F-117 CATIA V5 model as my starting point. However the model featured many details which would unnecessarily complicate a CFD simulation such as antennas, tiny features and especially fillets with very small radii (confuses the mesher) which had to be removed. The resulting model is significantly reconstructed and simplified. It was also converted to a solid in order to ensure full compatibility with SimScale. Finally, the model was cut in half to make use of the Symmetry condition and to cut down on computation times.

It was then imported to the SimScale project as a STEP file.

For the meshing, the SimScale snappyHexMesh algorithm was chosen. In order to ensure a properly developed flow, a base mesh box 10D upstream and 12D downstream buffer was created. Two cylindrical refinement regions were defined around the aircraft. Then, the surfaces were grouped and assigned refinements depending on their position, geometry and importance in affecting the flow. The refinements on the fuselage were defined between 6 and 9 levels.

The final mesh consists of 10.457.377 elements which is sufficiently fine to resolve the flow in detail.

Simulation

For the simulation part, I wanted to simulate the aircraft in flight. The following settings were chosen:

  • Analysis type: Fluid and compressible
  • Turbulence model: SST k-omega
  • Steady state (we’re not interested in time dependent features and we know we can achieve sufficient numerical convergence)

Flight and environmental settings:

  • Freestream velocity: Ma 0.4 (initially Ma 0.92, simulating cruise speed, later revised)
  • AoA: 5 degrees
  • Pressure: 100 kPa
  • Temperature: 5 degrees

However, ensuring a stable simulation turned out to be a bit tricky. After some investigation and various parameter revisions, this was finally achieved by using a smart trick used in a public project by SimScale staff Ali_Arafat. Simply, by using a CVS file, the inlet freestream velocity components were started out from very low values and slowly ramped up to their maximum values with each timestep. This ensured that the solvers didn’t have to deal with immense pressure and temperature levels in the initial timesteps which caused numerical instability.

The final simulation took 219 minutes on 32 cores and achieved good convergence. Ux residuals stayed high as during the first 750 timesteps while it was ramped up. It quickly started converging as soon as the ramping stopped.

Results

The results file was post-processed in ParaView.

The pressure distribution (Pressure range compressed to show relevant details) revealed highly undesirable aerodynamic properties as expected. There’s an unusually large high-pressure area in front of the cockpit and the engine intakes, which contributes to lift-loss, pressure drag and moment instability. Intake pressure efficiency of the engine should be affected by this as well.

At the sharp edges of the fuselage we see steep pressure drops followed by high pressure zones which indicate geometry induced separation as expected. This especially seems problematic around the intake and immediately afterwards we see a high-pressure zone which suggest massive flow detachment there. Same thing happens at the sharp edges on the upper wing surface, but the flow seems to reattach itself immediately afterwards.

I wanted to investigate the low-pressure area immediately behind the cockpit in the mid-fuselage-section as well, which suggests a large vortex formation and the resulting low-pressure area, possibly a clever trick utilized by Lockheed Skunkworks Engineers to increase lift by utilizing the vortex lift phenomenon.

The compressibility effects also follow the pressure distribution closely.

A simple streamline analysis confirms my prediction of the presence of a vortex at the top of the fuselage, behind the cockpit.

Finally, a quick drag and lift analysis. Theyalso follow the pressure distribution closely.

Conclusion

This was a very quick but effective analysis by using the SimScale platform. There are a lot of potential improvements that can be made to the simulation; including an even more precise and finer mesh, simulation of trans-sonic flight region, different AoA’s, perhaps a transient simulation to analys the vortex structures and of course; even more detailed post-processing in ParaView to extract as much information as possible from the simulation. I was especially very limited in that regard as my laptop could not handle many ParaView operations.

I hope you enjoyed reading about this little project of mine and I hope it inspires others to delve further into the wonderful world of CFD. I’d also like to thank to SimScale for this wonderful computing platform with a very user friendly web interface.

If you’re further interested about F-117’s aerodynamics, you can read the Master’s Thesis titled “Investigation of the F117A Vortical Flow Characteristics” by Sabine Anne Vermeersch.


#3

Hi, @emadarali That is Awesome. Great work ! :grinning: