Session: 01-05-01: Applications of Artificial Intelligence/Machine Learning for Aerospace Structures

Paper Number: 109144

109144 - Machine Learning Assisted Composite Rotor Blade Planform Design 

Composite materials are increasingly used in the design of rotor blades due to their high stiffness-to-weight ratio and excellent fatigue life. Currently, the composite rotor blade planform design intensively focuses on optimizing planform parameters to achieve higher performance. However, the strength of the material is rarely considered in the planform design. One issue is that the failure analysis of the rotor blade is usually performed by an expensive multiscale analysis which requires calculating the sectional properties, computing the rotor blade global response, and reproducing local stress for each layer of the composite blade. This process is inefficient when applying to an optimization framework, as whenever there is a new structural load, this analysis procedure needs to be performed again. Unfortunately, millions of load cases can be accumulated during the blade design optimization.
The ignoring of strength analysis may result in the blade working in an unsafe or low safety factor region, as composite materials are anisotropic and susceptible to various failure mechanisms. In this paper, we propose to optimize the composite rotor blade planform design with strength consideration. The optimized design can improve the performance of the aircraft and ensure the structure is within the safety margin. To reduce the computational cost of the cross-sectional failure analysis, we will use machine learning model to construct a beam-level failure criterion surrogate model to replace the physics-based cross-sectional failure analysis. The surrogate model is constructed based on the Timoshenko beam model via Artificial neural networks (ANN), where the mapping will be between blade loads and the strength ratios of the cross-section. Two optimization examples will be presented to demonstrate the capability of the proposed approach. 

Presenting Author: Fei Tao Dassault Systemes

Presenting Author Biography: Dr. Fei Tao is working at Dassault Systemes Simulia Corp as a Solution Consultant. He got his Ph.D. from Purdue University, Aeronautics and Astronautics Engineering. His expertise is in machine learning, finite element analysis, constitutive modeling, micromechanics, and mechanics of composite structures and materials.

Authors:

Fei Tao Dassault Systemes
Su Tian Purdue University
Haodong Du Purdue University
Wenbin Yu Purdue University