Session: 04-20-01: Design of Engineered Materials and Components for Additive Manufacturing

Paper Number: 150682

150682 - 3d Printed Lattice Materials for Impact Absorption: Uncertainty Quantification and Multi-Material Designs 

Lattice materials, characterized by cellular microstructured designs, are artificial structures that exhibit a high strength-to-weight ratio and excellent impact resistance. While the mechanics of such materials have been extensively studied in the literature, there has been little effort on performance assurance and understanding uncertainties in quantities of interest, resulting from limited manufacturing precision. Thsese include both geometric and material uncertainties. In addition, existing lattice structures are commonly fabricated by 3D printing of single phase materials, which limits the exploitable design space.

This work investigates the static and dynamic loading behaviors of 3D printable lattice designs with classic topologies, such as hexagonal honeycombs, and particularly focuses on the performance uncertainty quantification (UQ) aspect due to potential fabrication errors. Multi-material lattice designs are fabricated, analyzed, and experimentally characterized. The multi-material printing is achieved using a 3D printer with dual nozzles and a Cura slicer. Two thermoplastic elastomer (TPE) feedstocks are used to increase material contrast and expand the design space.

Statistical data are first collected from measured geometrical discrepancies between nominal designs and actual prints. Following the Monte Carlo approach, a low-fidelity reduced-order model is employed to simulate the imperfect designs and quantify performance uncertainty based on the measured geometrical distributions. The reduced order model is constructed with 2-noded generalized beam elements, which significantly reduce the needed computational cost and allow for a large number of Monte Carlo simulations. The uncertainty quantification analysis with low-fidelity reduced order models is then further refined using a set of high-fidelity finite element simulations. The imperfect samples are constructed in LS-Dyna with selected combinations of geometrical errors. To reduce the number of simulations needed while retaining convergence, we adopt the Taguchi orthogonal array, a highly fractional orthogonal design matrix, to select a reduced subset of combinations of multiple design parameters. With the high-fidelity error propagation results, a metamodel is constructed to approximate the design-performance relationship and derive the uncertainty range. The combined low- and high- fidelity approach exhibits an efficient uncertainty quantification process balancing the computational cost and accuracy.

Although the ideal/nominal designs demonstrate promising mechanical properties under impact loading, the uncertainty quantification analysis indicates that the actual printed samples may exhibit a wide range of performance, compromising the robustness of these designs. Consequently, we propose multi-material lattice designs with higher contrast material feedstocks, which not only broaden the design/property space but also reduce performance uncertainty while maintaining energy-absorbing capacity. This study highlights the necessity of performance uncertainty quantification for practical applications of lattice materials and provides an uncertainty-aware workflow for robust multi-material lattice designs.

Presenting Author: Weidi Wang University of Massachusetts Lowell

Presenting Author Biography: Weidi Wang is a scientific researcher with over 6 years of experience in developing mechanical metamaterials for wave controlling applications such as impact mitigation and source localization. He specializes in the modeling and designing of artificial micro-structured materials, using custom reduced order models and deep generative learning algorithms.

Authors:

Weidi Wang University of Massachusetts Lowell
Drupad Kadiyalabhavani University of Massachusetts Lowell
Seyedhassan Nikpour University of Massachusetts Lowell
Nathalia Diazarmas University of Massachusetts Lowell
Joey Mead University of Massachusetts Lowell
Alireza Amirkhizi University of Massachusetts Lowell