Banca de QUALIFICAÇÃO: PRINCE AZSEMBERGH NOGUEIRA DE CARVALHO

Uma banca de QUALIFICAÇÃO de MESTRADO foi cadastrada pelo programa.
DISCENTE : PRINCE AZSEMBERGH NOGUEIRA DE CARVALHO
DATA : 29/08/2019
HORA: 09:00
LOCAL: Sala 414 do CTEC - UFRN
TÍTULO:

MULTIOBJECTIVE OPTIMIZATION OF AERODYNAMIC PROFILES USING GENETIC ALGORITHM

PALAVRAS-CHAVES:

Genetic algorithm; airfoil; Matlab®; XFoil.

PÁGINAS: 73
GRANDE ÁREA: Engenharias
ÁREA: Engenharia Mecânica
RESUMO:

In the last decades the search for optimal aerodynamic profiles has increased in
the face of the growing demand for high performance aerodynamic components.
This search has been taking place in both aviation, energy and automotive sectors.
However, profile optimization requires a multiobjective approach since there is not
a single variable that optimizes the profile (in relation to its aerodynamic
coefficients) without affecting other structural characteristics such as its thickness
or curvature (bending). To streamline optimizations, artificial intelligence systems
are used, among which stand out the genetic algorithms, which mimic the natural
selection mechanism in the search for a set of variables that serve as a solution to
the problem. To characterize a profile, it is necessary to print a flow around it and
obtain the pressure curves to then obtain the desired coefficients of the profile. One
software that simulates the flow in the profile is XFOIL, this program virtualizes the
pressure distribution panels in the profile to obtain the pressure curves and then
the coefficients of the profile. In the present work we built a Matlab® algorithm that
could access XFOIL to characterize the initial profiles, classify them as to the ability
to reproduce other profiles for the next generations of the genetic algorithm and
work its geometry in order to optimize them in relation to them. to user-defined
multiobjectives. The algorithm was able to start with a population of 300 known
profiles, evolve and find an optimal profile in relation to the rest of the population.
The profile found has a 10% higher efficiency than the other profiles. The algorithm
was better relative to time, where it was able to perform the optimization with 18
hours, while the other works of optimization in genetic algorithm present about 300
hours with very similar computational capacity. It is suggested in the future to
check the optimal profile on CFD work.

MEMBROS DA BANCA:
Presidente - 1338331 - RAIMUNDO CARLOS SILVERIO FREIRE JUNIOR
Externo ao Programa - 1647050 - SANDI ITAMAR SCHAFER DE SOUZA
Interno - 1445637 - WALLACE MOREIRA BESSA