Automation of T-Spline based 3D High-Fidelity Isogeometric Analysis in Abaqus Xiang Ren a , Jim Lua a , Xiaodong Wei b , Yongjie Jessica Zhang b a Global Engineering and Materials, Inc., Princeton, NJ 08540, USA b Department of Mechanical Engineering, Carnegie Mellon University, Pittsburgh, PA 15213, USA Abstract: Isogeometric analysis (IGA) has shown its attractive feature recently for integrating a Finite Element Analysis (FEA) and Computer Aided Design (CAD) into a single unified process. For stress analysts, it is common practice to convert a spline/NURBS based CAD model to polynomial based finite element mesh for analysis. However, for complex geometries such as optimized 3D printing objects, it is known that smooth curved geometry cannot be exactly represented by a discretized finite element mesh. As a consequence, the simulation results can either be imprecise or costly due to the geometric misrepresentation or a larger number of degrees of freedom required to achieve the same accuracy. A T- spline based 3D IGA is developed by CMU to fill the gap. First, a 3D CAD geometry is automatically converted to a T-spline control mesh, and then it is converted to analysis suitable T-spline elements for a high fidelity analysis. With unified and higher order T-spline basis functions selected for the geometry representation, FEA is directly conducted on the smooth CAD geometry without loss of accuracy in retaining key design details. By taking advantage of the Bézier transformation, the 3D IGA solution module can be implemented in Abaqus via its user-defined elements (UELs). To accurately pose boundary conditions for an IGA solution domain from the specified ones at a physical domain, special algorithms are implemented for mapping the Dirichlet type boundary condition from a physical boundary to the control points. A suite of numerical examples are selected to demonstrate the accuracy and rate of convergence for the stress response prediction of complex 3D components. Keywords: T-Spline, Isogeometric Analysis, Bézier extraction, IGAFA 1. Introduction The concept of isogeometric analysis (IGA) presented by Hughes (2005) has paved a path towards a close integration of engineering design and computational analysis. The essential idea of IGA is to apply the same basis functions for the representation of geometry in Computer-Aided Design (CAD) and the approximation of field variables in Finite Element Analysis (FEA). Since a single geometry model can be utilized directly as the analysis model, it can avoid the labor-intensive mesh generation process required for analysis. In recent years, it has shown its great potential to significantly improve the efficiency of design- through-analysis cycle. IGA has shown its advantage over standard low-order finite elements in terms of solution per-degree-of-freedom accuracy via its application in academia problems including fluid mechanics and turbulence (Evans and Hughes 2013), solid and structure mechanics (Deng et al. 2015), fluid-structure interaction (Kamensky et al. 2015), phase-field modeling (Borden et al. 2014), contact mechanics (De Lorenzis et al. 2011 and 2014), and optimization (Kostas et al. 2015). The enhanced accuracy of IGA is partially due to the higher-order smoothness of the basis functions used. While significant progress was achieved in the last few years, the biggest challenge is the rapid, (semi-) automatic construction of geometric models suitable for analysis. Even for a shell structure of a complex geometry, it is still a time-consuming and challenging process to construct a baseline IGA model, and to ensure the model has the desired features including good parametrization, sufficient mesh density in the regions of interest, and, most importantly, analysis suitability. A manually driven partition can be used to push the limits of existing IGA technology to construct an analysis suitable model but it requires intimate familiarity