CRII: OAC: Improved Cyberinfrastructure Usage through High-Fidelity Isogeometric Volumetric Spline Model Generation

Brigham Young University, Provo UT

Investigators

Abstract

Physics-based scientific and engineering inquiry relies heavily on the use of computational models, including the finite element method. For particularly complex models, such as those used in automotive crash or those of national defense interest. Current techniques to achieve these analyses consume significant computational resources, approximate the geometries of intended objects for analysis, and do not leverage modern high-accuracy computational tools for the finite element method. Even though research has progressed to show that so-called isogeometric finite element methods are more accurate than traditional finite element methods for computational analysis, the scientific and engineering community lacks the tools necessary to generate computational models suitable for such analyses. This research creates a framework through which scientists and engineers can convert a computer-aided model or mesh of a geometry or potential design into a three-dimensional representation that directly represents the intended domain without approximation and that leverages spline-based isogeometric tools for more accurate physics-based simulations. This work supports national security interests by providing access to higher-fidelity analysis results and by streamlining the process by which engineers arrive at these results. It also helps develop more accurate models for use by the automotive industry such as in simulating crash, resulting in safer and more sustainable vehicles. The developed toolset is made publicly available for use and continued enhancement. The research helps train and diversify the US cyberinfrastructure community through mentoring of undergraduate, graduate, and underrepresented groups in STEM fields. In particular, this work defines a technique to create a well-structured hexahedral decomposition of a computer-aided design geometry or a surface mesh that can be output for use either using traditional finite element methods or more structured isogeometric methods. The framework relies on concepts from differential geometry and Morse theory, and mathematically guarantees a valid volumetric discretization of the geometry from a surface parameterization. Both the theory and the computational tools constructed from the theory are validated by reconstructing (a) a vehicle of interest to the United States Army and (b) a left ventricle of a patient-specific heart model. Finally, both academic and industrial communities are provided access to the developed software through a permissive license that invites use and future development. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.