Project leader: Anders Logg
Principal investigator: Mats G.
Larson
The flow solvers are based on the software
developed by the Computational
Middleware Project and used as the basic building blocks for
simulating complex flow problems in the Biomedical Flows
Project. The key to building such robust solvers is a
posteriori error estimation techniques and adaptive algorithms that
automatically tune the local resolution in the individual solvers to
achieve overall accuracy.
As part of the project, we have developed
a framework for a posteriori error estimation and adaptive
algorithms for coupled multiphysics solvers. The framework enables
quantification of the influence of errors in an individual solver on a
given user-provided output quantity. Applications investigated so far
include coupled flow-transport problems and thermoelasticity. We are
currently extending the framework to fluid-structure interaction.
In another activity tied to this project,
we investigate how to apply the reduced basis method to the modeling of
hierarchical flow systems. Together with external collaborators, the
geometry of the computational domain has been introduced as a parameter
to allow efficient computation of flow in a series of pipes and
bifurcations by coupling precomputed basis functions on subproblems.
The long term goal is to apply the methodology to the full
three-dimensional, time-dependent Navier-Stokes equations and
fluid-structure interaction problems encountered in biomedical
flows.
Current activities
