Additive manufacturing produces spatially heterogeneous material states shaped by process history, microstructure, surface condition, residual stress, and populations of internal defects. This session will focus on computational methods that explicitly represent these sources of heterogeneity and connect them across length scales to deformation, fatigue, fracture, and component-level performance. Particular emphasis will be placed on models that move beyond nominal or homogenized material properties by incorporating measured or statistically reconstructed as-built features, quantifying variability and uncertainty, and establishing meaningful links between simulations and experimental observations. The session seeks contributions that clarify when computational models can provide predictive insight, how model fidelity should be matched to the engineering question, and how defect-sensitive simulations can support qualification, design, and life assessment of additively manufactured materials and components.

Scope and topics of interest

• Multiscale modeling linking processing history, microstructure, defects, and mechanical performance
• Explicit and statistical representation of pores, lack-of-fusion defects, inclusions, rough surfaces, and other as-built discontinuities
• Image-based modeling using X-ray computed tomography, microscopy, surface metrology, or other characterization data
• Crystal plasticity, phase-field, cohesive-zone, extended finite-element, and peridynamic models for AM materials
• Computational treatment of defect–microstructure and defect–surface interactions
• Modeling of crack initiation, short-crack behavior, fatigue life, and fracture from realistic defect populations
• Residual-stress and anisotropy effects on local deformation and failure
• Representative-volume approaches, synthetic microstructures, and statistically equivalent defect fields
  Reduced-order and surrogate models derived from high-fidelity simulations
• Uncertainty quantification, stochastic simulation, sensitivity analysis, and probabilistic life prediction
• Verification, validation, model calibration, and transferability across machines, builds, geometries, and loading conditions
• Integration of computational predictions with in situ testing and multimodal experimental characterization
• Computationally informed qualification, allowables development, component design, and fitness-for-service assessment
• Open modeling benchmarks, reproducible workflows, and shared AM datasets