SC²S Colloquium - July 8, 2015

From Sccswiki
Revision as of 11:53, 1 July 2015 by Bakhtiar (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search
Date: July 8, 2015
Room: 02.07.023
Time: 3:00 pm, s.t.

Nicolai Schaffroth: Simulation of rain-induced floods on high performance computers

To allow the simulation of rain-induced floods based on the shallow water equations, several additions and modifications were made to the SWE code base and the HLLE solver. The code, which was originally designed for tsunami modeling, was adapted for better precision at wet-dry interfaces and low water heights. The algorithms and data structures were updated to account for new parameters in the scenarios, including rainfall and friction. Several test scenarios and comparisons with benchmark solutions demonstrate the capabilities and limitations of the HLLE solver implementation.

Ionut-Gabriel Farcas: Uncertainty Quantification in Fluid-Structure Interaction

Fluid-Structure Interaction (FSI) problems, where one or more structures interact with an internal or surrounding fluid flow, play prominent role in many scientific fields, that pose many challenges do to their strong non-linearity and multidisciplinary nature. Hence, FSI is a complex domain by itself. Representative examples are blood flow simulations, flutter analysis of airplanes, wind turbines. Among the challenges in FSI is uncertainty, which appears naturally in most prob- lems (e.g in density of fluid or/and material, viscosity of the fluid, etc). Using forward- propagation of the uncertainty in input parameters, the aim of this work is to investigate the uncertainty in FSI problems, via sophisticated mathematical methods and high per- formance computing. In more details, non-intrusive methods based on sampling will be used. These methods are Monte Carlo sampling (MCS) and Stochastic Collocations (SC), that is based on generalized Polynomial Chaos (gPC) methodology. Due to the complex nature of FSI, the UQ methodology will be implemented on certain model FSI problems. Lastly, multi-dimensional stochastic problems are also being considered. Whereas MCS is independent of the dimensionality (and of the input pdf), in SC the curse of dimensionality occurs. In order to overcome it, sparse grids will be employed and to this extent, the SG++ library is to be used.