Hammering Away at the Dynamic Stall Problem to Build Better Rotorcraft
05.23.19
This novel visualization shows—for the first time—blade vortex interaction (BVI) causing dynamic stall on a Black Hawk helicopter rotor in forward flight. Notice the green vortex passing over the rotor blade, which causes the flow to separate (shown in red) rather than smoothly passing over the blade. When this happens, the rotor blade loses thrust and pitches downward, causing significant vehicle vibrations and limiting the maximum flight speed. Neal Chaderjian, Tim Sandstrom, NASA/Ames
For more than ten years, NASA research scientist Neal Chaderjian has been developing improved simulation software that can be used to design safer, faster, and more energy-efficient rotorcraft. Improving aerodynamic performance during rotorcraft vertical takeoff and landing plays a critical role for medical and military transport, firefighting, police surveillance, and the emerging urban air mobility (UAM) market. Chaderjian, who works at the NASA Advanced Supercomputing (NAS) facility at Ames Research Center, has improved our fundamental understanding of rotorcraft flows, including dynamic stall.
It is challenging to accurately predict dynamic stall, where air flow separation on a helicopter rotor causes a sudden decrease in thrust and large structural vibrations—ultimately limiting the vehicle’s maximum flight speed. “And everyone wants to go faster,” Chaderjian notes.
I’m building better computational tools so that engineers can build a better rotorcraft.
Neal Chaderjian, NASA Ames
To help rotorcraft engineers and designers better understand this phenomenon, Chaderjian has developed increasingly accurate computational fluid dynamics (CFD) flow simulations, and provides guidance on how to accurately simulate the flow at reduced computational cost. He uses NASA supercomputers, combined with the NAS Division’s visualization expertise, to generate simulation results that are freely shared with the rotorcraft and CFD communities. Along the way, he has also made some first-time observations and discoveries that are influencing CFD technologies and aeronautics research worldwide.
This is the best part of his work, Chaderjian said—gaining new insight into complex rotorcraft flows and then providing accurate flow simulation technology to help design engineers build safer, more efficient aircraft in less time and for less money. “It’s like developing a better hammer so the carpenter can build a better home—I’m building better computational tools so that engineers can build a better rotorcraft.”
Tackling Dynamic Stall Simulation Challenges
New faster designs eventually encounter dynamic stall, Chaderjian said. For example, a helicopter pilot can increase flight speed by increasing the rotor’s thrust and tilting the vehicle forward, but at some point the rotor can’t deliver any more thrust and stalls, preventing the vehicle from flying faster.
Dynamic stall is also very challenging to accurately simulate, even with advanced CFD tools. Many rotorcraft engineers think about dynamic stall as a mostly two dimensional phenomenon, which neglects important three-dimensional effects such as blade vortex interaction (BVI)—where the vortices of rotor blades interact with each other. Chaderjian's simulations show that BVI can actually trigger dynamic stall, something completely neglected in the two-dimensional model.
So, in addition to performance calculations, Chaderjian’s high-fidelity CFD model explores the fine details of the turbulent vortical flow in order to provide greater insight into the fundamental causes and nature of dynamic stall. Fortunately, his toolbox includes NASA’s powerful OVERFLOW Navier-Stokes CFD code, which allows him to dynamically refine the computational grid on the blade surface and in the rotor wake where it’s most needed.
In a recent study of a Sikorsky UH-60 Black Hawk helicopter, Chaderjian applied OVERFLOW’s near-body adaptive mesh refinement (NB-AMR) to the dynamic stall problem. This was the first time NB-AMR was applied to a flexible rotor blade in forward flight and showed BVI caused dynamic stall under the right flow conditions. According to Chaderjian, just a year ago it was thought that only the blade’s high angle of attack caused dynamic stall. He also showed that the CFD simulation was “grid converged,” meaning no further grid refinement will improve the accuracy of the simulation. His conclusion: further improvements in accuracy will likely require better measurement of the blade structural properties.
It’s important for CFD experts like Chaderjian to provide design engineers guidance on how to best use CFD methods with confidence. Part of what keeps Chaderjian pounding away at these issues is curiosity: “I want to know what’s causing what,” he said. The CFD community has been long aware of how difficult it is to predict and control dynamic stall. In this case, Chaderjian found that BVI can be an important mechanism that triggers the phenomenon.
Tools of the Trade: Supercomputing and Visualization
Chaderjian’s high-resolution Black Hawk rotor simulations required thousands of grids and more than a billion grids cells per case. Solution times varied from 40–180 hours using 5,600 Skylake cores on the NAS facility’s powerful Electra supercomputer. He ran these simulations at the highest resolution possible and long enough to be confident the solution was as accurate as possible. From that point, he was able to back off and see what minimal resolution and run times would provide the same accurate answer. While savings in computer time can vary substantially depending on the problem being solved, overall, users can expect a savings of 2–10 times—or more if AMR is used.
Yet Chaderjian knows it’s not enough to simply look at plots and graphs. That’s where the importance of working closely with the NAS Division’s scientific visualization team comes in.
NAS visualization expert Tim Sandstrom provided the first animations that revealed the dynamic nature of the flow simulations, which gave Chaderjian the idea that BVI could be triggering dynamic stall. But the complex flow “covered up” what was happening near the blade surface. Sandstrom came up with an innovative approach that provided a “window” to see through all of the flow to the blade surface. From these visualizations, Chaderjian developed his theory on how BVI can cause dynamic stall and then validated his theory with an additional simulation.
“The movies are more than pretty pictures—they are the key to helping us understand what causes the flow to behave the way it does,” Chaderjian said.
Blueprints for Better Rotorcraft Flow
Many design engineers don’t have access to large supercomputing facilities. Chaderjian’s ability to run many high-resolution simulations on Pleiades and Electra has allowed him to make progress toward his long-term goal: to obtain the most accurate solutions, using the least number of computing cycles, in the shortest amount of time.
Chaderjian is now working on plans to further improve the efficiency and robustness of the NB-AMR algorithm that will enable design engineers to utilize CFD tools with greater confidence and reliability, and introduce CFD sooner into the design process to shorten the design time and costs.
“Rotorcraft are very complex vehicles that give rise to flow phenomena and interactions that often take years of study to understand. This also motivates and drives the development of more accurate simulation tools,” Chaderjian said. But that’s part of the fun, so he keeps hammering away.
—Jill Dunbar, NASA/Ames Research Center
One of the goals of NASA’s Revolutionary Vertical Lift Technology (RVLT) Project is to develop and validate advanced high-fidelity computational fluid dynamic (CFD) tools with improved physical realism and numerical solution accuracy that can be used in the rotorcraft design process.
Related Links
Papers
- Navier-Stokes Simulation of UH-60A Rotor/Wake Interaction Using Adaptive Mesh Refinement, Chaderjian, N. M., presented at the American Helicopter Society 73rd Annual Forum, Fort Worth, TX, May 9-11, 2017. (PDF 8.2 MB)
- Numerical Simulation of Dynamic Stall Using Near-Body Adaptive Mesh Refinement, Chaderjian, N. M., Paper Number ICCFD10-102, presented at the 10th International Conference on Computational Fluid Dynamics, Barcelona, Spain, July 9-13, 2018. (PDF 8.1 MB)
