Beyond lattices: A geometry-first computational design framework for additive manufacturing of porous architectures

Additive manufacturing (AM) has enabled unprecedented geometric freedom in architected porous materials, yet computational design workflows remain largely based on periodic, feature-driven strategies in which predefined unit cells are assembled and optimized to achieve target properties. While effective, this approach struggles to represent the multiscale heterogeneity, disorder, and scale coupling observed in natural and high-performance porous systems. Recent developments increasingly adopt stochastic, field-based, and generative architectures, often without a unifying conceptual framework. This Perspective introduces a chaos-driven, geometry-first approach in which porous architectures emerge from underlying deterministic nonlinear dynamical rules, while features such as pore size, connectivity, and anisotropy are extracted a posteriori as descriptive metrics. Within this framework, additive manufacturing is reinterpreted as a scale-selection and truncation mechanism determining which levels of multiscale geometry are physically realized. The approach remains physically informed, as nonlinear dynamical systems can reproduce multiscale and coupled behaviours relevant to material and transport phenomena. The resulting architectures resemble natural structures such as wood, not through direct imitation but through the implicit reproduction of underlying generative processes, thereby inverting the conventional topology-driven design paradigm.