
Resolving Cross-Talk in Immersed Isogeometric Shell Analysis : A Comparative Study of Selective Control Point Duplication and Adaptive Local Refinement
Please login to view abstract download link
Immersed isogeometric analysis is a numerical approach that combines the high accuracy and high continuity offered by isogeometric analysis (IGA) [1] with the modeling simplicity offered by immersed boundary methods [2]. As opposed to boundary-fitted approaches, immersed IGA embeds the geometry into a non-conforming background mesh and accounts for the actual geometric boundaries through dedicated numerical integration schemes. However, the modeling of small geometric discontinuities with immersed methods can be inaccurate due to cross-talk. That is, small trimming features in an immersed model, such as thin notches, small holes, or cracks, may cause diverse unphysical behavior. This includes for instance force transmission through the void domain and stress oscillations near the trimming features. Some efforts have been made in understanding and tackling cross-talk in immersed IGA, shells in particular. Coradello et al. [3] showed that cross-talk can be asymptotically eliminated through local h-refinement using truncated hierarchical B-splines (THB-splines). Lian et al. [4] provided a mathematical definition of cross-talk and classified the corresponding control points into two types. Type1 cross-talk refers to the control points with disjoint support domains, whereas Type2 cross-talk corresponds to those whose support are connected but non-convex. Furthermore, in [4], an alternative solution named control point duplication (CPD) was proposed which effectively eliminates Type1 cross-talk while preserving the stable time step size in explicit dynamic analysis. In this contribution, we present (i) an a-priori cross-talk potential estimation framework; (ii) an extended CPD approach that also handles Type2 and hybrid types of cross-talk; (iii) adaptivity of CPD and local refinement; (iv) a study that compares selective CPD to various adaptive local refinement strategies.