Physiological fluid dynamics generally involves the interaction of a
viscous incompressible fluid with a visco-elastic biological tissue. The
Immersed Boundary Method provides a robust framework for the computer simulation
of biofluid dynamic systems, even when the configuration of the biological
tissue is complicated, dynamic, and not known in advance. Although the
Immersed Boundary Method is able to furnish qualitatively good results,
it suffers from a certain "lack of resolution'", not related to the method
itself but to limitations of computers such as speed and storage. Usually
thin boundary layers develop along the immersed biological tissue, requiring
a dense computational mesh to resolve adequately the flow there. If a uniform
mesh is used, this requirement is inevitably extended to the entire computational
domain, and the resulting mesh may exceed the storage capacity of the computer.
The aim of this work is to present a new computational setting for the
Immersed Boundary Method, in which the resolution can be enhanced locally
by using the Adaptive Mesh Refinement Technique. In this new setting a
vicinity of the immersed boundary is covered by a sequence of nested, progressively
finer rectangular grid patches. This strategy concentrates the computational
effort for the regions of the flow where it is most needed.
Also, an implicit formulation of the method is introduced to free the
scheme from the time step restriction coming from the stiffness of the
immersed boundary. The numerical results show that with this new setting
the method succeeds in achieving the same accuracy as if the whole computation
had been performed on an uniform mesh with the resolution of the finest
grid patches that are used to cover the immersed boundary. Potential applications
of the methodology presented include the computer assisted design of artificial
heart valves and other biological problems where high resolution is needed
in a neighborhood of the immersed elastic boundary to capture the boundary
layers which are formed there.
Roma, A.M.: A multilevel self adaptive version of the Immersed Boundary Method. Ph.D. Thesis, CIMS-NYU, 01/1996. University Microfilms #9621828.