LIFE SCIENCES
Simulating Transitional Hemodynamics in Intracranial Aneurysms at Extreme Scale
Principal Investigator:
Sabine Roller
Affiliation:
Simulation Techniques and Scientific Computing, University of Siegen (Germany)
Local Project ID:
pr85mu
HPC Platform used:
SuperMUC of LRZ
Date published:
Scientists leverage high performance computing technologies to identify the morphological characteristics of intracranial aneurysms that result in high frequency fluctuations, and assess the role of these fluctuations in aneurysmal wall degradation and consequently aneurysm rupture. Using SuperMUC they performed simulations with up to one billion elements, which allowed the simulation of flow at spatial and temporal resolutions of 8µm and 1µs, while resolving the smallest structures that can develop in a turbulent flow.
Stroke caused by the rupture of intracranial aneurysms (IA) remains a major cause of morbidity and mortality in the modern world. Nearly 4-5% of the world population is reported to be suffering from un-ruptured IA. Computational fluid dynamics (CFD) is a promising tool to assess the factors that lead to initiation, growth and rupture of an aneurysm by computing the hemodynamic forces like wall shear stress and pressure.
The intra-aneurysmal flow, due to slow movement of the blood, is usually considered laminar. With the advent of modern supercomputers however, it became possible to not only demonstrate but also quantify the intricate hemodynamics in aneurysms that exhibit high-frequency fluctuations resembling a weakly turbulent or transitional flow.
The primary goal of this research is to identify the morphological characteristics of aneurysms that result in such high frequency fluctuations, and assess the role of these fluctuations in aneurysmal wall degradation. The open source lattice Boltzmann solver Musubi, developed at the institute of Simulation Techniques and Scientific Computing of the University of Siegen, ran several setups for 24 hours on 32,000 cores of HPC system SuperMUC of LRZ to perform simulations with up to one billion elements. The use of SuperMUC has allowed us the scientists to simulate flow at spatial and temporal resolutions of 8µm and 1µs, while resolving the smallest structures that can develop in a turbulent flow.
Figure 1: Peak systolic flow characteristics in an aneurysm. Miniature vortices educed by velocity colored Q-iso-surfaces (left), volume rendering of Vorticity (middle) and velocity magnitude overlapped with velocity vectors (right).
Copyright: Simulation Techniques and Scientific Computing, University of Siegen (Germany)The researchers’ on-going efforts include taking heart rate variability into account for simulations and its potential role in the transitional hemodynamics. Moreover, they are investigating the role of surrounding vasculature on aneurysmal hemodynamics that increases the scale of simulations to 2 billion cells.
The scientific team consists of the following researchers:
Kartik Jain1,2, Harald Klimach1, Sabine Roller1 (PI), Kent-Andre Mardal2,3
1 Simulation Techniques and Scientific Computing, University of Siegen, Siegen, GERMANY
2 Center for Biomedical Computing, Simula Research Laboratory, Lysaker, NORWAY
3 Department of Mathematics, University of Oslo, Oslo, NORWAY
Acknowledgements:
The simulations were performed on the GCS supercomputer SuperMUC installed at the Leibniz Supercomputing Center, Munich (Grant – pr85mu).
Scientific Contact:
Prof. Dr.-Ing Sabine Roller
Simulation Techniques and Scientific Computing, University of Siegen
Hölderlinstr. 3, D-57076 Siegen/Germany
E-mail: sabine.roller@uni-siegen.de
LRZ project ID: pr85mu
October 2015