Computing physiological flows: From cutting meshes and automated code generation to optimization of flow patterns

Speaker: Andre Massing
Affiliation: Simula Research Laboratory, Norway
When: Tuesday 26th November 2013
Time: 10:00:00
Abstract: In recent years, the field of virtual medicine has started to investigate simulation-based analysis as a tool to predict the impact of medical interventions or to investigate the design of medical devices such as stents and coils. Starting from the underlying model problems for two medical conditions, namely cerebral aneurysms and Chiari I malformation, we highlight specific challenges in the pipeline for simulating blood and cerebrospinal fluid flows. Emphasis will be put on the automatization of the simulation pipeline and the possibility for a seamless integration of various pre-/post-processing steps and solver routines into a single simulation framework. Here,the FEniCS project plays an essential role to provide a flexible and versatile high-level finite element framework for automated and efficient solution of partial differential equations. We illustrate > how the mathematical expressiveness and automated code generation allows rapid implementation, prototyping and testing of the solution algorithms in questions without compromising efficiency. Building upon FEniCS and its finite element library DOLFIN, the dolfin-adjoint framework provides a powerful tool to perform sensitivity analysis, data assimilation and optimization in an automatized manner. Finally, we focus on recent finite element methods on cut meshes (CutFEM). CutFEM technologies allow flexible representations of complex or rapidly changing geometries by decomposing the computational domain into several, possibly overlapping domains. Alternatively, complex geometries only described by some surface representation can easily be embedded into a structured background mesh. We demonstrate how CutFEM techniques can be employed to address various challenges from mesh generation to fluid-structure interaction problems and optimization tasks. ref: