November 18, 2009
Small Aircraft Will Mimic Dragonfly
CFD visualization used at Wright State University to study the complex flow physics of dragonfly flight
Most students caught hanging out in the school parking lot during class might find themselves in a bit of trouble, but for Dr. Haibo Dong's grad students, it's a study requirement."They've been out there all morning trying to catch a few dragonflies," he laughs. "We have a high-speed camera in the lab, but don't have any subjects to film. Catching them can be hard because they're so fast and maneuver so well. But that's why we're studying them, isn't it?"
The camera equipment comes from one of several multi-year grants awarded to a team of researchers led by Dr. Dong at Wright State University to study dragonfly flight. Their long-term goal is to develop quad-winged micro-air vehicles (MAVs) that mimic the flight of a dragonfly and could be used to monitor activity in public places, like sports stadiums or public transit stations, to help detect biological or chemical weapons. Such vehicles also can be used to search for victims or fires, bombs, and other threats in spaces that are unsafe or too small for people to enter.
What makes Dr. Dong's project so exciting is that it is the first to create full-body, 3D simulations of a flying four-winged insect using a time-sequence of control surfaces reconstructed from high-speed images. Researchers have long studied two-winged insects such as bees and flies in their efforts to develop bi-wing aircraft. They've also known that quad-winged aircraft modeled after dragonflies would enjoy more lift, better maneuverability, and greater speed. But the complexities of the insect's body motion, wing deformation and interaction, and the subsequent flow physics have been too difficult to unravel.
"Until recently, scientists have only been able to look at individual insect parts. They could look at a modeled wing in motion, for example, but not consider the wing deformation and the multi-wing interactions," he says. "We are particularly interested in the exact full-body 3D simulations to study the flow physics of a four-winged insect like the dragonfly. We need to see how its wings and body deform or flex in relation to each other as it moves through the air. We need to look at the vortices, how they form, how they pass over the wings. We need to see how they respond to different conditions in nature. And all of it must be extremely precise."
Building the computer model
It has been a slow, painstaking process for Dr. Dong's team because of the large amount of data involved and the need for precision. The first step was to construct an efficient simulation tool and then validate it against hundreds of scientific publications, studies, and observations on insect motion.
It took his team more than three years to collect and validate the CFD data needed to describe how insects take off, maneuver, cruise, hover, and an almost endless number of additional nuanced wing movements. Finally, in early 2009, they were ready to take the next step: creating the 3D simulations with enough precision, detail, and accuracy to recreate every motion a dragonfly makes on computer. Zach Gaston, an undergraduate student, began by using three high-speed cameras at different angles to film dragonflies taking off. Chris Koehler, a computer science Ph.D candidate, used advanced computer vision algorithms to track points in the video images and AutoDesk Maya to reconstruct the exact 3D shape of both body and wings based on these points. He then pulled the resulting 3D kinematics into Tecplot 360 where they were able to start running CFD simulations of the flight. Next, Zongxian Liang, a mechanical engineering Ph.D candidate, used in-house simulation tools to conduct direct numerical simulation of the flow using the reconstructed motion.*
"The visual results showed the exact motion of a dragonfly taking off, including its body motion and wing deformations," says Dr. Dong. "Now, by varying our validated mathematical model, we can begin to understand with great confidence how a dragonfly manages its wings to create the forces needed to lift its body up under a variety of conditions. Next, we need to repeat the process for other stages of flight like hovering, maneuvering, and landing."
While the 3D animations grab the most attention and fire the imagination, Dr. Dong's team also relies heavily on 2D images, plots, and charts created with Tecplot 360 to analyze their data. Still images help them examine the action frame by frame. XY plotting allows the team to look at force history to learn how the wing is generating force or where the air is moving. 2D cross-sections can offer more detail on leading edge vortices, for example, that show how they influence force production.
"The 3D animations are probably the coolest part. We can rotate the body, look at the structure, examine vortices, and gain a deep understanding of the physics," says Dr. Dong. "But this is physics-based analysis, and that always involves lots of plots and graphs. To us, the information we glean from the 2D results is just as exciting."
When will we actually see quad-winged MAVs in use? Dr. Dong's team expects to deliver their first proto-type sometime next year, but a true commercial application is still five to seven years out.
"Five years isn't that far off and there are a lot of people out there who want to make it happen," he says. "For example, we've received solid cooperation from scientists at Wright Patterson Air Force Base. They see the potential this technology offers in helping protect the public's safety."
A welcome side effect: providing inspiration for future scientists
An unexpected benefit of the Tecplot 360 images and animations is that they have made a powerful recruiting tool for future scientists. Dong and his fellow researchers regularly demonstrate their research and results to high school students who visit their campus, and the animations and images helps spark their young imaginations.
"This past summer, for example, we sponsored a camp for high school girls to encourage them to become scientists or engineers. They were absolutely fascinated by these animations." says Dong. "It's so much easier for kids to get excited when they see a visual of a bio-inspired problem like this. You show them data and, well, it's just a bunch of numbers to them. But convert the data into an animation? They see it, they get it. It really makes them want to learn the intricacies of science."
* Other project contributors include Dr. Hui Wan and Matt Maples from the Mechanical Engineering Department and Dr. Thomas Wischgoll in the Computer Sciences Department at Wright State University.
Images Courtesy Tecplot
On the Net: