Optimization of Composite Structures for Crashworthiness

Abstract

Structural optimization for crashworthiness in composite structures has become an important topic of research in aerospace, attributable to its proven benefits to the occupants' safety in an aircraft during a crash event. This thesis provides a comprehensive investigation on LS-DYNA modeling and design optimization of composite square tubes for crashworthiness. The objective of this thesis is to construct a optimization framework specialized for the preliminary design of composite structures with a balance of performance and efficiency. Three primary components are involved: firstly, coupon-level simulations in LS-DYNA are performed to characterize the material properties for carbon/epoxy composite material IM7/8552; secondly, a square tube is modeled by a single-layer approach in three mainstream material cards (MAT-54, MAT-58 and MAT-262) with calibrated parameters for crush simulation in LS-DYNA. Meanwhile, detailed sensitivity analysis of influential parameters in different material models is also be performed to have a more comprehensive understanding of the complex failure mechanism. The simulation results indicate good correlation to the experiments in terms of energy absorption and maximum peak load, with high computational efficiency and low-cost calibration. Lastly, a two-stage single-objective optimization is performed, which incorporates the fiber orientation for each layer as design variables and design/manufacturing rules as constraints. Two surrogate models are created to formulate the mapping between input design variables and output crashworthiness metrics, including Deep Neural Networks (DNN) and Gradient Boosting Regression Trees (GBRT) ensemble. Followed by Mix-Discrete Particle Swarm Optimization (MDPSO) algorithm, the optimal set of design variables are obtained for each surrogate model. The first-stage optimization results demonstrate significant improvement in the crashworthiness performance compared with the baseline value, while the second-stage optimization results indicate an excellent transferability of the proposed optimization framework. This applicable, transferable, and data-driven optimization framework can be used in the aerospace industry regarding the crashworthy design and optimization of composite structures.