Text

theses

Advancements in aerospace technologies which rely on unsteady fluid dynamics are being hindered by a lack of easy to use, computationally efficient unsteady CFD software. Flapping in nature is ubiquitous, yet modern day micro air vehicles (MAVs) based on flapping are in their infancy. The most successful MAVs to date are much less maneuverable and efficient than their natural counterparts, partly due to the fact that they are based on the relatively simple aerodynamics of propeller blades. Similarly, mainstream wind turbines are based on rotary blades and relatively simple aerodynamics. In fact, flapping wing energy harvesters have been increasingly investigated as a possible alternative to traditional wind and tidal turbines after several studies highlighted their unique capabilities and exceptional efficiencies.
Flapping wing energy harvesting has been shown in recent years to be reaching efficiency levels “comparable to the best performances achievable with modern rotor blade turbines” [1]. Simultaneously, advancements in experimental and computational abilities are bringing a comprehensive understanding of the underlying mechanisms of flapping wing insect flight within reach. These mechanisms may hold the key to highly maneuverable and resilient MAVs [2], and some of them have been shown to enhance the performance of flapping foil power generators as well [3].
The major reason for the scarcity of flapping devices is the difficulty involved in the design of such devices. Existing CFD platforms are capable of handling unsteady flapping, but the time, money, and expertise required to run even a basic flapping simulation makes design iteration and optimization prohibitively expensive for the average researcher. For the design of a MAV capable of the complex maneuvers observed by natural flyers, this lack of computational efficiency makes the successful implementation of a viable design nearly impossible.
However, by utilizing a novel unsteady vortex method which has been designed specifically to handle the highly unsteady flapping wing problem, it has been shown [4] that the time to compute a solution is reduced by a factor of 20, and the level of skill to operate the software is reduced so much that an undergraduate engineering student can easily produce accurate results.
Despite the success of the original vortex method from [4], especially for a small number of flapping cycles, the solution deteriorates as the number of flapping cycles increases due to the inherent lack of viscosity in the vortex method. This would make it challenging to couple the fluid solver to a rigid body solver, as the increasing moment would cause the MAV to tumble. In addition, the original vortex method from [4] does not utilize parallel processing to increase the computational speed and wastes a large amount of computational time in the far field wake for simulations involving multiple flapping cycles. Finally, the original method assumed the motion of the MAV to be fully prescribed, whereas a more valuable tool for MAV designers would allow for the motion of the MAV body to be predicted concurrently with the fluid forces given a specified wing motion.
It is therefore the goal of the present study is to create a faster, suitably accurate fluid-rigid body interaction simulation for flapping wing MAV designers. This is accomplished by further improving the computational efficiency of the solver and by adding the capability for the simulation to solve for the flight trajectory and aerodynamic forces of a flapping wing MAV concurrently given only the flapping kinematics. This simulation thereby allows for testing virtual flight stability, maneuver flapping sequences, and performance testing.
First, the code will be shown to run orders of magnitude faster by being modified to allow the GPU to compute vortex velocity contributions in a massively parallel configuration.
In addition, a remedy which models the effect of viscosity is introduced into the original vortex method. The new approach proposed herein lumps far field vortices to simulate viscosity-induced vortex decay, which will be shown to improve the accuracy of the solution while maintaining the pitching moment amplitude. This is especially important for simulations involving many flapping cycles, which is the case when predicting the flight path of an MAV. In addition to improving the accuracy of the solution, the new method greatly reduces the computation time for simulations involving many flapping cycles. Several different incident flow velocity angles were tested and the moment amplitude is shown not to increase for all cases.
Moreover, a novel fluid-rigid body interaction simulation is shown to leverage the improvements to the fluid model to allow for the equations of motion of a two-body flapping wing flyer to be solved. This new fluid-rigid body solver is then utilized to support the hypothesis that the position of a flapping wing insect’s abdomen is carefully adjusted in order to balance the pitching moment created by aerodynamic forces generated from flapping. Additionally, a basic control feedback loop is introduced to simulate an MAV’s on-board active flight stabilization control system. This simulated control system is shown to stabilize the MAV’s main body, thereby creating a stable platform for mounting sensors, such as a camera. Finally, the solution of the original vortex method from [4] and the solution of the present method are compared to published data from a full Navier Stokes simulation [5] and show good agreement.

ETD_9756

Ph.D.

Includes bibliographical references

doi:10.7282/t3-e3f6-2065

ETD doctoral

The author owns the copyright to this work.