Sequential parameter estimation for fluid-structure problems: Application to hemodynamics

We present a robust and computationally efficient parameter estimation strategy for fluidstructure interaction problems. The method is based on a filtering algorithm restricted to the parameter space, known as the reduced-order unscented Kalman filter. It does not require any adjoint or tangent problems. In addition, it can easily be run in parallel, which is of great interest in fluidstructure problems where the computational cost of the forward simulation is already a challenge in itself. We illustrate our methodology with the estimation of the artery wall stiffness from the wall displacement measurements as they could be extracted from medical imaging in a three-dimensional idealized abdominal aortic aneurysm. We also show preliminary results about the estimation of the proximal Windkessel resistance, which is an important parameter for setting appropriate fluid boundary conditions. Copyright (c) 2011 John Wiley & Sons, Ltd.