Furkan Onal

Middle East Techinal University, Turkey

Abstract Title:Prediction of Unsteady Flow Field Around Missile Geometry by Using Deep Learning Methods

Biography:

M. Furkan ÖNAL has completed his B.Sc in 2019 from Middle East Technical University Aerospace Engineering department and continues M.Sc studies from same university and department. He is the aerodynamics specialist at Roketsan Missiles Launch Vehicle Design department for 4 years. His main interests are aeroacoustics, aerodynamics, fluid dynamics and artificial intelligence.

Research Interest:

The main purpose of this study is predicting the unsteady flow field around a body by using multi-layer perceptron (MLP) deep learning method. To train a reliable artificial intelligence model; three-dimensional, steady state, Euler based computational fluid dynamics (CFD) analyzes have been performed with wide range of angle of attacks and Mach numbers. The simulations have been performed with SU2 open source CFD solver. The results have been used as the training dataset to feed and train the MLP model. The model trained with steady-state results with varying Mach numbers and angle of attacks. Afterwards, unsteady dynamic motion predictions have been performed for different reduced frequency values. Two quantities which are Mach number distribution around geometry and pressure distribution on geometry surface have been predicted very well up to high reduced frequency values. The results indicate that this method works very well and can speed up preliminary design phases of missile projects in industry.