Anisotropic phase-field fracture parameters: Evolutionary algorithm perspective

Polylactic acid (PLA) plays a prominent role in medical implants, packaging, and the textile industry, among the various industrial sectors. Components can be efficiently 3D printed by the Fusion Deposition Modeling (FDM) process, which however is inducing a material anisotropy due to the layer-by-layer deposition. The phase field (PF) approach to fracture generalized to handle anisotropic brittle materials is herein critically examined since it offers potential capabilities to simulate crack paths in such materials. Since the formulation is based on an anisotropic structural tensor ω with the incorporation of penalty parameter β , this governs the material fracture energy G c , the internal length scale l c , and the apparent strength. The novel contribution of the work lies in integrating a metaheuristic machine learning algorithm (MLA) with the PF approach to robustly estimate fracture parameters ( G c , l c and β ) and get an insight into epistemic uncertainty of the formulation. Results highlight that particle swarm optimization (PSO) is robust in estimating fracture parameters to reproduce target force-displacement response curves. Sensitivity analysis of fracture parameters reveals the critical role of β in influencing fracture predictions. • Meta-heuristic algorithms applied to anisotropic phase-field for fracture. • Sensitivity analysis of the phase-field parameters. • Identification of phase-field fracture parameters. • Most efficient metaheuristic machine learning algorithm identified. • Epistemic uncertainty.