Thompson, David S.
Riveros, Guillermo A.
Luke, Edward A.
Date of Degree
Dissertation - Open Access
Computational Engineering (Program)
Doctor of Philosophy
James Worth Bagley College of Engineering
Bio-structures owe their remarkable mechanical properties to their hierarchical geometrical arrangement as well as heterogeneous material properties. This dissertation presents an integrated, interdisciplinary approach that employs computational mechanics combined with flow network analysis to gain fundamental insights into the failure mechanisms of high performance, light-weight, structured composites by examining the stress flow patterns formed in the nascent stages of loading for the rostrum of the paddlefish. The data required for the flow network analysis was generated from the finite element analysis of the rostrum. The flow network was weighted based on the parameter of interest, which is stress in the current study. The changing kinematics of the structural system was provided as input to the algorithm that computes the minimum-cut of the flow network. The proposed approach was verified using two classical problems – three- and four-point bending of a simply-supported concrete beam. The current study also addresses the methodology used to prepare data in an appropriate format for a seamless transition from finite element binary database files to the abstract mathematical domain needed for the network flow analysis. A robust, platform-independent procedure was developed that efficiently handles the large datasets produced by the finite element simulations. Results from computational mechanics using Abaqus and complex network analysis are presented. The complex network strategy successfully identified failure mechanisms in the bio-structure by identifying strain localization in regions of tension, and buckling/crushing in regions of compression. The transdisciplinary strategy used in this study identified the failure mechanisms early, when the material was still in the linearly elastic regime, thereby tremendously reducing the computational time and cost as compared to running a finite element analysis to failure. This work also developed five proof-of-concept, bio-inspired models with varying lattice complexity based on the rostrum. Performance of these bio-inspired models was analyzed with respect to the stress and deformation. Numerical experiments were carried out on one of the bio-inspired model to demonstrate the application of newly developed similitude laws for blast loading. This research has laid the groundwork for an efficient design-test-build cycle for rapid prototyping of novel bio-inspired structures by using flow network analysis, finite element analysis, and similitude laws.
Patel, Reena R, "Complex Network Analysis for Early Detection of Failure Mechanisms in Resilient Bio-Structures" (2018). Theses and Dissertations MSU. 1126.