Uncertainty Quantification in Multi-Phase Cloud Cavitation Collapse Using Multi-Level Monte Carlo
Speaker: Jonas Šukys, Computational Science & Engineering Lab
Presentation
Seminar video is unavailable due to technical difficulties.
Abstract
Cloud cavitation collapse pertains to the erosion of liquid-fuel injectors, hydropower turbines, ship propellers, and is harnessed in shock wave lithotripsy. To assess the development of extreme pressure spots and other key quantities of interest for such systems, we performed numerical two-phase flow simulations of cloud cavitation collapse containing up to 50,000 bubbles using up to 1 trillion domain discretization cells. Initial system conditions such as pressure ratio, geometrical distribution and sizes of the bubble cavities are prone to uncertainty, and were observed to have a strong influence on the magnitude and position of the peak pressure. To quantify the propagation of aforementioned uncertainties in collapsing cavity clouds under uncertain configurations, we employ a novel non-intrusive, multi-level Monte Carlo methodology, aiming to accelerate the plain Monte Carlo sampling. Highly heterogeneous confidence intervals obtained at different stages of a collapse process provide valuable insight for further analysis of this volatile phenomenon.
Biography
Jonas Šukys is a postdoctoral fellow at the Computational Science & Engineering Lab in Zürich, Switzerland. His research interests focus on Uncertainty Quantification, multi-level Monte Carlo methods, hyperbolic nonlinear stochastic partial differential equations, multi-phase flows and cavitation dynamics, and massively parallel high performance computing.