Multi-Layered Randomized Architected Materials (MLRAM) as damage detection indicators for tensegrity structures
Advisor: Prof. Ann C. Sychterz
Abstract
Emergence of additive manufacturing and advanced 3D printing technologies has enabled the design and fabrication of novel material architectures, accelerating research in the field of architected materials and metamaterials. This thesis introduces a new type of architected material inspired by polymeric structures and explores two central research themes. The first is whether data-driven models can accurately and efficiently predict material properties, offering computational and time advantages over traditional numerical approaches, while the second examines whether the proposed architected material can serve as a visual indicator of damage for tensegrity structures. Four primary objectives are developed to perform the research presented in this thesis.
The first objective focuses on the design and fabrication of a novel two-layered architected material, called the Multi-Layered Randomized Architected Material (MLRAM), inspired by polymeric geometries and produced using 3D printing technology. The fabricated MLRAM specimens were experimentally tested under monotonic tensile loading to evaluate their mechanical behavior.
The second objective involves the development and automation of a finite element method (FEM)-based numerical model capable of accurately replicating the tensile response of MLRAM specimens. An automated simulation framework, integrating MATLAB and Abaqus, was established to efficiently generate and analyze the force-displacement behavior of 2,400 unique MLRAM configurations.
The third objective utilizes the dataset generated from the numerical simulations to develop data-driven predictive models using machine learning and deep learning algorithms, enabling the estimation of key material properties.
The fourth objective involves the integration of a specifically designed MLRAM element into a tensegrity structure to demonstrate its structural applicability and functional performance as a damage detection indicator. A prototype of a tensegrity hollow-rope footbridge, comprising four pentagonal tensegrity-ring modules, was developed in the laboratory. Tensegrity systems consist of struts and cables, where struts can sustain both tension and compression, whereas cables carry only tensile loads. In the prototype, a critical cable was identified based on stress distribution under externally applied loads. An MLRAM element was specifically designed to match the design tensile capacity of this critical member and integrated into the cable. Under static loading, as stresses in the bridge members increased, the incorporated MLRAM failed at its design limit, providing a clear visual indication of localized damage.
Results demonstrate that randomization of the MLRAM geometry successfully replicates the macroscopic characteristics of polymeric structures and yields a wide range of tensile load capacities. This allows for the selection of specific designs tailored to desired performance criteria. Automating the numerical simulation process enabled efficient analysis of large specimen sets with minimal human intervention. Machine learning and artificial neural network models achieved satisfactory accuracy in predicting the peak tensile load capacity of the MLRAM specimens. Experimental validation confirmed that MLRAM failure coincides with the critical cable reaching its design load, offering a visible indicator of structural distress.
Findings from the primary objectives helps in drawing conclusions towards the central research themes. Combining experimental investigations with numerical modeling can generate extensive datasets essential for developing robust data-driven prediction tools. Data-based models demonstrated a significant reduction in computation time compared to traditional numerical simulations, highlighting their potential for future engineering applications. Furthermore, the successful integration of MLRAM as a visual damage detection element in tensegrity structures introduces a concept of sensor-free visual health monitoring, potentially reducing dependency on conventional sensing technologies.
Keywords: Architected material, Artificial intelligence, Tensegrity structures, Damage detection