Development of a Computational and Experimental Framework for Characterizing Defects in Additively Manufactured Lattices

Abstract

Additive manufacturing (AM) enables the production of complex, architected lattice structures with tunable mechanical properties. However, the inherent process variability and unique defects associated with AM introduce uncertainties in part quality, frequently causing premature failure and inhibiting broader industrial adoption. Given that not all defects render a part unfit for end-use, there is a critical need for scalable metrological frameworks capable of characterizing and qualifying defect-laden AM lattice parts. This research presents a computational and experimental framework to systematically design, manufacture, and analyze defect-engineered AM lattices. The framework programmatically generates dozens of unique lattice designs with stochastically injected defects of controlled magnitude and location. We demonstrate the efficacy of this framework by manufacturing 762 physical samples across three polymer resins (Formlabs Rigid10k, Tough 1500, and BlackV4). Custom automated metrology software rigorously analyzes the as-manufactured lattice quality by extracting precise thickness, length, orientation, and straightness data for every element of each lattice sample. Finally, the framework’s utility is validated through a mechanical testing case study on 330 of the Rigid10k and Tough 1500 samples. While baseline performance is predominantly dictated by nominal geometric dimensions and bulk material properties, the pipeline successfully enables isolation of specific factors that degrade compressive strength and stiffness. We find that missing elements and the overall defectivity ratio contribute most significantly to the reduction in mechanical properties and may serve as critical stress concentrators. By establishing a scalable pipeline connecting programmatic defect generation, automated metrology, and empirical mechanical testing, this work provides the foundational infrastructure required to develop robust qualification standards and predictive performance models for AM lattices.