On the Sedimentation of a Blob of Particles Suspended in a Viscous Fluid
- Wilber Cerqueira ,
- Rafael Gabler Gontijo ,
- Sara Malvar
23rd ABCM International Congress of Mechanical Sciences and Engineering |
Published by ABCM
The main goal of this article is to identify the vibrational patterns of an oscillating spherical bubble immersed in
a ferrofluid using Neural Networks. The gas bubble is immersed on a magnetic fluid and subjected to an harmonic pressure
excitation and an applied magnetic field. The classical Rayleigh-Plesset equation is modified and a magnetic version is
proposed in order to model the radial motion of the bubble. Being a very non-linear system, the correct identification of
the vibrational patterns is not possible using linear system tools. In this case, fuzzy logic and neural networks are a good
solution for the correct pattern identification. The magnetic version of Rayleigh-Plesset equation is a second order non
linear ordinary differential equation and it is solved through a fifth order Runge-Kutta scheme with adaptive time step.
Four different vibrational patterns are identified and proposed regarding its frequency response and Lyapunov exponets.
Based on the Lyapunov exponents it is possible to analyze its non-chaotic behavior. Those Lyapunov exponents are used
in a backpropagation neural network to correctly identify the bubble vibration modes what could lead to the identification
of the flow physical parameters in practical applications. In order to compare the performance, another neural network
is trainned with the amplitude and frequency information of the bubble motion from ω= 0 to ω= 10, based on its Fast
Fourier Transform. The new results are compared with previous simulations made with harmonic response in frequency
spectrum, in which only the first harmonic was considered