Eventos Anais de eventos
ENCIT 2020
18th Brazilian Congress of Thermal Sciences and Engineering
NEURAL NETWORK ALGORITHM FOR PREDICTION OF AERODYNAMIC COEFFICIENTS OF A REDUCED SCALE ROCKET
Submission Author:
Rodrigo Daher , MG , Brazil
Co-Authors:
Laura Castro, Rodrigo Daher, Alexandre Zuquete Guarato
Presenter: Rodrigo Daher
doi://10.26678/ABCM.ENCIT2020.CIT20-0806
Abstract
Computational Fluid Dynamics (CFD) simulations are extremely important in the fluid mechanics field, as they allow to obtain aerodynamic data without having the need to perform experimental simulations. On the other hand, machine learning, in a simplistic way, allows the computer to understand the behavioral connections between the inputs and outputs. The Aerospace Technology and Propulsion Team (EPTA) design and manufactures reduced scale rockets for competitions from Federal University of Uberlândia. This paper aims to present a neural network algorithm developed to predict the rockets aerodynamic data, so it would be possible to easily obtain the aerodynamic data and simulate the trajectory of the rocket without the need of experimental launches. In this paper CFD simulations were performed on Ansys software and the neural network was developed thought a python algorithm. Aerodynamic data from the proposed algorithm have been compared to a similar rocket model using RASAero II software, which is based on Barrowman’s method to compute the aerodynamic coefficients. Results from the proposed neural network algorithm have a maximum 17.74 % deviation in relation to Ansys and RASAero II, which is accurate enough to be used in the team.
Keywords
CFD, Aerodynamics, machine learning, neural networks
DOWNLOAD PDF VIEW PRESENTATION
