• Efficient mesh morphing for complex turbine-blade geometries from X-ray CT data. • Robust CAD-to-CT surface matching via functional maps and Laplace-Beltrami operator. • Regularized CAD-mesh morphing preserving labels, connectivity and quality. • Automatic transfer of PDE boundary conditions from CAD to CT meshes. • FE thermal fields and lifetime predictions compare nominal, CT and morphed blades. We propose an automatic mesh-morphing framework for industrial digital twins, enabling high-fidelity simulations on as-manufactured geometries of high-pressure turbine blades. X-ray computed tomography (CT) captures the real geometry, and a reference CAD finite-element mesh is morphed toward the scanned part while preserving mesh identity, namely node ordering, element connectivity, and associated labels and sets. This preservation enables robust registration and automatic transfer of boundary conditions, regions, and load definitions from the nominal CAD model to the CT-derived geometry. The approach combines functional maps with the Laplace-Beltrami operator to compute reliable correspondences between nominal and scanned surfaces, overcoming the limitations of closest-point projection on thin-walled industrial parts. The resulting surface morphing is then extended to the volumetric mesh while maintaining mesh quality and keeping surface-defined boundary conditions consistent. The industrial relevance is demonstrated on a real turbine blade through comparative thermo-mechanical finite element simulations performed on the nominal CAD model, the morphed CAD mesh, and the CT-reconstructed geometry, complemented by life assessment computations. Results show that incorporating as-manufactured geometries into the simulation chain remains compatible with automation constraints and supports rigorous high-fidelity prediction in industrial workflows.
Ferhat et al. (Wed,) studied this question.