Session: 01-04-02: Advances in Aerospace Structures

Paper Number: 107302

107302 - An Efficient High-Fidelity Design Tool for Advanced Tailorable Composites 

Affordable space explorations and innovative aircraft concepts require lightweight structural solutions with possible complex shapes. The use of composite materials is the de facto solution to lightweighting of aerospace structures such as hybrid/blended wing body, space launch vehicles, space habitats, etc. Most existing composites designs are limited to quasi-isotropic laminates made of unidirectional fiber-reinforced composites (UDFRCs) and cannot harness the full potential of advanced composite materials to produce highly tailorable designs with optimized load paths. Tailorable composites reinforced with long or short fibers have been proposed for lightweight aerospace structures for enhanced performance. However, there is no commercially available design tool for the design and analysis of advanced tailorable composite structures. This is because existing design tools were created especially for composite laminates consisting of straight fibers. Compared with UDFRC structures, the design and analysis of advanced tailorable composites possess the following features: 1) UDFRCs consist of straight and continuous fibers, while tailorable composites consist of continuous and/or discontinuous fibers whose orientations vary by location and result in location-dependent stiffness/strength. 2) The composite modeling modules of most existing commercial finite element tools are designed for UDFRCs and cannot not directly define location-dependent structural properties. 3) The fiber paths in advanced tailorable composites can be described by a series of design variables, making it possible to optimize structures under a complex loading and/or environmental conditions.

These features motivate us to develop a new efficient high-fidelity Design tool for Advanced Tailorable Composites (DATC), including the following components: a) Mechanics of structure genome (MSG): efficient high-fidelity modeling of advanced tailorable composites i) capable of accurately predicting stiffness and strength of general composites featuring arbitrary anisotropy and heterogeneity, and ii) fully compatible with the standard structural elements in commercial FEA packages so that real structures made of advanced tailorable composites can be routinely designed and analyzed; b) A versatile parameterization method i) capable of expanding the design space for optimization of advanced tailorable composites, ii) considering varying fiber orientations, ply coverages (varying ply thickness and ply -drops), different materials, and manufacturing constraints, and iii) utilizing general-purpose optimizers capable of producing highly tailorable, optimal designs; c) An integrated design framework i) providing user-friendly GUI plug-ins for MSC.Patran/Nastran and Abaqus, ii) exploiting MSC.Nastran and Abaqus’s versatile capabilities in design and analysis of advanced tailorable composites, iii) leveraging all MDAO capabilities (parameter study, sensitivity analysis, optimization, UQ, etc.), and iv) capable of being integrated into other broadly used engineering codes such as HyperWorks, HyperSizer, and LSOPT; d) A machine learning (ML) module integrated with the GUI plug-ins for developing efficient surrogate models i) capable of training ML models from the plug-ins without expert ML knowledge and extra programming efforts, ii) capable of reproducing MSG-based constitutive modeling with significantly reduced computing time, and iii) capable of reproducing the entire analysis including constitutive modeling with orders of magnitude savings in computing time.

Presenting Author: Su Tian Purdue University

Presenting Author Biography: PhD student, Aeronautical and Astronautical Engineering, Purdue University,
Purdue University, Aeronautical and Astronautical Engineering, MS, 2015
Tongji University, China, Aerospace Engineering, BS, 2013

Authors:

Su Tian Purdue University
Xin Liu University of Texas at Arlington
Liang Zhang AnalySwift
Wenbin Yu Purdue University