Embedded Boundary Methods for CFD and FSI Problems: Challenges and Solution


Embedded/Immersed Boundary Methods (EBMs/IBMs) are very attractive for CFD on non body-fitted meshes. This is because they introduce a high degree of automation in mesh generation, and a significant flexibility in meshing complex geometries. In the presence of rough surface textures such as those encountered, for example, in the geometrical descriptions of heat shields of re-entry vehicles, iced aircraft, feather-covered bird wings, and submarine hulls with biofouling, the highly automated nature of meshing for EBMs is a significant advantage over body-fitted meshing. EBMs also continue to gain popularity for the solution of highly nonlinear Fluid-Structure Interaction (FSI) problems with topological changes. For such applications, alternative methods based on Lagrangian or Arbitrary Lagrangian-Eulerian approaches are either cumbersome, or simply unfeasible. Nonetheless, EBMs suffer some serious disadvantages. To begin, the solutions and solution gradients they deliver at a wall boundary typically suffer from accuracy issues and can be sensitive to the position and orientation of the embedded surfaces with respect to the embedding mesh. This is partly because EBMs work in general with a jagged surrogate of a material interface, and partly because they introduce ill-conditioning in the spatial discretization. Consequently, EBMs do not predict many quantities of interest such as flow separation points and the wall skin friction as accurately as body-fitted CFD methods. This talk will discuss all of these issues and will present proven approaches for resolving them. It will also illustrate these approaches with the solution of several challenging CFD and FSI problems. These include: the simulation of a turbulent flow problem over a bird wing characterized by a feather-induced surface roughness; the prediction of the thrust generated by a complex flexible flapping wing as a function of the flapping frequency, and the correlation of the obtained numerical results with experimental data; and the solution of a multi-scale parachute inflation dynamic problem associated with NASA's Low Density Supersonic Decelerators project.