Preprint Open Access

Aerodynamic Reduced-order Volterra Model of an Ornithopter under High-amplitude Flapping.

C. Ruiz; J.Á. Acosta; A. Ollero

The unsteady aerodynamics of apping low-aspect-ratio ellipsoidal-wings in ornithopters is analyzed
and modeled by the use of three dimensional Computational Fluid Dynamics (CFD) simulations.
The range of interest is large amplitude, moderate frequency apping, and low to moderate angles of
attack at Reynolds around 10^5, where autonomous ornithopters like GRIFFIN, are able to perform
complex maneuvers such as perching.
The results obtained show that the Leading Edge Vortex is produced above a certain Strouhal and
angle of attack at downstroke. These aerodynamic loads are compared with the classical analytical
models by the frequency response, observing that analytical models based on abscence of viscosity and
small perturbations are not appropriate for the range of interest as the hypotheses are not fulfilled.
Through the 3D CFD aerodynamic loads database, a finite memory Volterra model is identified in
order to predict the characteristics of forces and moments produced by the apping wing. This reduced
order model depends on the efective angle of attack of the surrogate airfoil located at 70 % of the
semi-span at three-quarters chord on the airfoil. This state has been found appropriate for being
the one with the greatest regression, comparing the 3D CFD simulations with others that have been
carried out in 2D, in agreement with the literature.
Finally, a methodology to validate the identified model without the need of wind tunnel is proposed
and validated for lift force. By the use of the aerodynamic forces extracted from ight data, measured
by a high accuracy Motion Capture System at diferent apping wing kinematics, it is concluded that
the model provides better estimates than classical analytical models.
The structure of the model and its predictability make it possible to use it in control tuning, in
addition to being able to append nonlinear or aeroelastic terms using a similar method, also because
of the execution time, provide a potential solution for online forces prediction.

Files (17.6 MB)
Name Size
17.6 MB Download
All versions This version
Views 8383
Downloads 7979
Data volume 1.4 GB1.4 GB
Unique views 7070
Unique downloads 7171


Cite as