High-fidelity prediction of sintering deformation in DLP-fabricated Ti6Al4V components via inverse calibration of a viscous sintering model

Digital light processing (DLP)-based fabrication offers a promising route to producing high-resolution Ti6Al4V components, but severe and anisotropic shrinkage during sintering remains a major challenge for dimensional accuracy, particularly in complex geometries. Accurate prediction is hindered by the high computational cost and numerical difficulty associated with constitutive parameter calibration in thermomechanically coupled sintering simulations. In this study, a prediction framework based on the Skorohod-Olevsky viscous sintering (SOVS) model is combined with a sampling-driven inverse calibration strategy using Progressive Latin Hypercube Sampling (PLHS) and a kinematic-volumetric constraint for constitutive parameter identification. The calibrated model is validated over a wide sintering temperature range and shows good agreement with experiments, with prediction errors below 3% for relative density and anisotropic shrinkage. For Schwarz P, Gyroid, and gear structures, geometry-dependent deformation is predicted with dimensional deviations below 4% from macro- to micro-scale features, while residual stress analysis reveals topology-dependent stress localisation in complex architectures. These results demonstrate that the proposed framework provides a practical approach for dimensional compensation and near-net-shape manufacturing of DLP-fabricated Ti6Al4V components.