Parallel multiscale simulations of a brain aneurysm.

Area of Science:

• Computational fluid dynamics
• Molecular dynamics
• Biomedical engineering
• Cardiovascular research

Background:

• Cardiovascular pathologies like brain aneurysms are influenced by both global blood flow and local microrheology.
• Modeling these conditions requires integrating continuum (PDEs) and atomistic (ODEs) scales, posing significant mathematical and computational challenges.
• Accurate simulation necessitates bridging disparate spatial and temporal scales in computational models.

Purpose of the Study:

• To present a novel hybrid methodology for multi-scale simulations of platelet deposition on brain aneurysm walls.
• To enable the first simulations coupling continuum and atomistic approaches for this specific biomedical problem.
• To investigate clot formation within patient-specific aneurysm models.

Main Methods:

• Coupled spectral element Navier-Stokes solver for large-scale intracranial flow and dissipative particle dynamics (LAMMPS) for blood microrheology.
• Interface conditions using adaptively computed effective forces ensure continuity between continuum and atomistic domains.
• Immersed boundary method tracks evolving platelet clusters; a multilevel message passing interface links heterogeneous solvers.

Main Results:

• Successfully performed the first multi-scale simulations of platelet deposition and clot formation in a brain aneurysm.
• Demonstrated a two-way interaction between deposited platelets and the blood flow model.
• Presented scalability results of the coupled solver up to 300,000 processors.

Conclusions:

• The developed hybrid methodology effectively couples continuum and atomistic simulations for complex cardiovascular problems.
• This approach provides a powerful tool for studying clot formation in patient-specific aneurysm geometries.
• Further validation of coupled atomistic-continuum models remains a critical area for future research.