Building a robotic hummingbird is probably not the first thing one associates with Army research, but bird-sized robotic aerial vehicles are much in demand. Micro aerial vehicles, or MAVs, could perform surveillance, hazardous materials detection, terrain mapping, or data relay functions in situations where it would be hazardous or impractical to send humans. MAVs don’t need sleep, and they don’t get bored by looking at the same terrain for hours at a stretch. Unmanned aerial vehicles (UAVs), larger counterparts to the MAVs, have proven useful for military operations in Afghanistan, but the current generation of vehicles requires a runway for takeoff, and they are unable to hover.

AHPCRC researchers from New Mexico State University are working on a robotic hummingbird wing to gain insight into how to mimic a hummingbird’s maneuverability. They have incorporated information from videos, provided by Bret Tobalske (then at the University of Portland, now at the University of Montana), of the vortices that live hummingbirds make as they fly. Hummingbird wings flap and twist, creating eddy currents that give the bird lift and momentum.

Mingjun Wei and Banavara Shashikanth, NMSU assistant professors of mechanical and aerospace engineering, head up the project. The late James Allen, NMSU assistant professor of mechanical engineering, oversaw the project in its early stages. Prof. Tom Burton, department head, was instrumental in bringing the NMSU group into the AHPCRC program, and he participates in the research group meetings. Prof. Charbel Farhat of Stanford University is providing additional computer simulation work on lift and drag. NMSU graduate students Tao Yang, Humberto Bocanegra Evans, Mohammad Mazharul Islam, and Ramiro Chavez receive AHPCRC support.

The research team is building computer simulations that will help them understand and imitate the complex wing motions that come naturally for small birds and insects. At the scale of a small bird, you have to deal with factors that don’t even register for large aircraft (or even large birds)—a gust of air from a building’s exhaust vent, or drag generated by the viscosity of the air, for example. Birds and insects have been dealing with wind gusts, fuel intake and consumption, lift and drag, and obstacle avoidance issues for millennia. They fly under all sorts of weather conditions, alone or in flocks or swarms. (See previous article for more on small bird flight.)

How does a bird do it?
Among the problems an MAV designer has to face is how to gather and store enough energy to sustain flight without making the vehicle too heavy to fly. The hummingbird does this by feeding every 10–15 minutes during its waking hours and slipping into a deep sleep state at night. It would be impractical to operate an MAV on this schedule, so efficient energy usage is key to the success of these vehicles.

Birds and insects make very little noise as they fly, and they recover quickly if they graze a tree branch or other obstacle. These will be important features in designing MAVs, along with agility and maneuverability. Hummingbirds are especially agile—they can take off from a stationary position, hover, and change directions in midair. During flight, they sweep their wings forwards and backwards while plunging (oscillating) and pitching (rotating) their wings, tracing figure-eight paths with their wing tips.

Building a better bird
The earthbound engineers at NMSU are using high-performance computer (HPC) simulations to explore the best combinations of plunging, pitching, and twisting motions for producing thrust. They are testing various frequencies and amplitudes of the flapping and twisting motions to determine what works best. They are developing a three-dimensional HPC version of a code based on a modified immersed boundary method (see page 5), which models a flowing fluid interacting with a flexible, elastic structure (such as a wing) by computing the flow and the structure simultaneously on the same mesh grid in a fully-coupled fashion. Dynamical models for flow–structure interactions are being developed for some simple configurations. Wei and Yang are doing numerical simulations, and Shashikanth and Islam are developing theoretical models.

Evans, Chavez, and their co-workers have built a model wing that they are testing in a low-turbulence water channel, and they will use their findings to validate and improve their computer models. This will enable them to find useful configurations, maybe even ones that Nature hasn’t thought of yet, and save their mechanical model-building resources for the best candidates. Preliminary runs have been conducted on a mechanical model wing with two degrees of freedom of motion, and the concept design has been completed for a wing with three degrees of freedom.

Farhat’s group at Stanford is pursuing an alternative simulation approach based on a combination of higher-order ALE (Arbitrary Lagrangian Eulerian) and embedded methods. This approach is especially useful for modeling the interactions of moving flexible wings with air streams, because it enables calculations based on the interaction of both solid and fluid reference frames.

To this effect, they have developed an upgraded version of the massively parallel AERO-F computer code that features a number of explicit and implicit schemes that satisfy their discrete geometric conservation laws (DGCLs)—a set of conditions imposed on a moving computational grid in order to preserve the accuracy and stability of the calculations when integrated over a period of time. The formal accuracy of these schemes was proven to be identical to their counterparts on fixed grids. The team has analyzed the amount of thrust generated by high-frequency plunging motions using AERO-F ’s large eddy simulation (LES) capabilities. More recently, they simulated flapping motions (active plunging coupled with passive pitching) for various frequencies.

In particular, LES runs were performed at Stanford on massively parallel processors to investigate numerically the existence of a critical frequency, below which drag is generated and above which thrust is produced. The simulations clearly demonstrated the existence of such a critical frequency. Previous experimental studies had shown a surprising onset of nonsymmetric wake patterns at higher frequencies. The Stanford simulations also predicted these patterns, which produce both lift and thrust.

Insights gained through computational modeling and simulation provide valuable guidance in choosing configurations and operational parameters for real-world mechanical wings, and for understanding the factors that go into making a wing that works. Back at NMSU, work continues on improving the physical wing models for testing in the water channel. The research group also plans to build a computational model of a micro aerial vehicle with flapping wings. They will test their models using plunging, pitching, and twisting motions alone and in combination to determine which motions produce the most thrust.

Source: AHPCRC Bulletin, Vol. 1, Issue 3 (2009)