Paper Submission
01. Experimental/Computational Fluid Dynamics
EVALUATING THE EFFICIENCY OF PINNS IN SIMULATING NACA 4 SERIES AIRFOILS
In the past, Computational Fluid Dynamics (CFD) was used to simulate the coefficients of lift and drag forces to improve aircraft lift and efficiency. However, the wide variety of airfoils available made traditional CFD an impractical choice due to its time-consuming nature. In recent years, artificial intelligence (AI) has emerged as a promising alternative for numerical prediction. Currently, Physics-Informed Neural Networks (PINNs) are being utilized to assist in establishing flow field models. This study aims to develop a model for the NACA 4 series airfoil through a single training session.
By altering three main parameters—the maximum camber, the location of the maximum camber, and the maximum thickness of the wings—researchers can compare the impacts of these different parameters on performance. To validate the model's accuracy, we will compare the predictions to results obtained from CFD. Despite significant differences at the trailing edge, PINNs remain a valuable tool for identifying performance trends, even when minor shortcomings are present.
While the previous section highlights some advantages of PINNs, they also have inherent issues, such as increasing aircraft lift without considering the risk of stall. Although PINNs are highly beneficial for design purposes, maintaining accuracy and accelerating computations remains a significant challenge.
Download the file you uploaded
Author Information
M. Chang
Ms.
Corresponding author, Presenting author
H. Chang
Mr.
W. Wang
Dr.