Reliability-Based Design of Additively Manufactured Michell Structures for Aerospace Certification

Abstract The widespread adoption of additive manufacturing (AM) in safety-critical aerospace structures is hindered by process-induced variability and the limitations of traditional deterministic certification methods. This report presents a comprehensive, multi-fidelity, uncertainty-aware framework for the reliability-based design and certification-oriented assessment of topologically optimized structures. Drawing inspiration from Michell's theory of minimum-weight truss layouts, a series of prototypes were fabricated via fused deposition modeling (FDM) and tested under uniaxial quasi-static loading. The framework was demonstrated through testing of Michell structures that supported loads exceeding 200 N at masses below 30 g under quasi-static conditions. Physical experimentation was combined with high-fidelity finite element modeling, calibrated through Bayesian inference to account for material anisotropy and process variability. To ensure computational efficiency during Bayesian calibration and Markov Chain Monte Carlo (MCMC) sampling, neural network–based surrogate models were developed. Sensitivity-driven dimensionality reduction was applied to isolate dominant uncertainty drivers and improve computational efficiency within the stochastic simulation pipeline. The framework quantifies AM-induced variability and demonstrates an end-to-end digital substantiation process that integrates verification and validation (V&V) and data traceability. Results confirm the viability of combining simulation and experimental evidence to compute reliability indices and assess design acceptability under uncertainty. This work contributes to the advancement of model-based aerospace qualification by bridging topology optimization, probabilistic modeling, and regulatory alignment.