Avatar

Alexander Jaust

Postdoctoral researcher

University of Stuttgart

Biography

I am a postdoctoral research in the group of Prof. Miriam Schulte at Stuttgart University at the Institute for Parallel and Distributed Systems (IPVS). Here I am member of the collaborative research center SFB1313 about “Interface-Driven Multi-Field Processes in Porous Media - Flow, Transport and Deformation”. My work within the SFB is concerned with efficient partitioned coupling schemes for porous media applications.

Before working in Stuttgart I obtained a master’s degree in Computational Engineering Science at RWTH Aachen University (Germany). After my master’s I started a PhD at RWTH Aachen University and Hasselt University (Belgium) where I was part of a joint PhD program. During my PhD studies I mainly focused on efficient high-order methods for computational fluid dynamics. I moved to the supercomputing center at KU Leuven where I worked as high-performance computing analyst/consultant for a year before going to Stuttgart.

You can find my current hompage at the Institute for Parallel and Distributed Systems here .

Interests

  • Partitioned coupling methods for porous media applications
  • High-order methods in computational fluid dynamics
  • High-performance computing

Education

  • PhD in Computational Mathematics, 2018

    Hasselt University and RWTH Aachen University

  • MSc in Computational Engineering Science, 2013

    RWTH Aachen University

  • BSc in Computational Engineering Science, 2011

    RWTH Aachen University

Experience

 
 
 
 
 

Postdoc

University of Stuttgart

Dec 2018 – Present Germany
Postdoctoral researcher at the Institute for Parallel and Distributed Systems in the group of Prof. Dr. Miriam Schulte. I am a member of the SFB1313 and work on partitioned coupling schemes for porous media applications.
 
 
 
 
 

HPC Analyst/consultant

Hasselt University and KU Leuven

Dec 2017 – Nov 2018 Belgium
High-performance computing analyst/consultant supporting users with using the supercomputers at the Flemish supercomputer center (VSC). My tasks ranged from helpdesk support to the implemention of new features as the integration of a JupterHub service running on the supercomputer.
 
 
 
 
 

PhD student

Hasselt University and RWTH Aachen University

Feb 2014 – Oct 2018 Belgium and Germany
PhD student with Prof. Dr. Jochen Schütz (UHasselt) and Prof. Dr. Manuel Torrilhon (RWTH Aachen) where I was part of a joint PhD program. I did research in the field of high-order discretizations for computational fluid dynamics.

Recent & Upcoming Talks

Coupling Free and Porous-Media Flow via Partitioned Black-Box Methods

Many applications involve flow in porous media and some other medium that can be separated into subdomains by a sharp interface. One example is coupled surface and subsurface flow. Simulating such applications poses several practical challenges such as, e.g., the condition number of the monolithic linear system of equations or the use of different mesh sizes in the different subdomains. We investigate a coupling using a partitioned coupling scheme using the open-source library preCICE (https://www.precice.org/) avoiding the typical problems of monolithic couplings. It follows a black-box approach which allows for a simple coupling of different disretizations or even different software packages for the subproblems while avoiding the ill-conditioning of a monolithic approach.

Simulation of fractured porous media using partitioned black-box methods

The simulation of coupled fracture flow and deforming porous medium is a challenging problem in reservoir engineering. Common examples are hydraulic simulations or hydro-fracking. Some of the challenges arise due to the difference in properties of the mathematical models used in each of the subdomains. Solving the problem using a monolithic approach leads to an ill-conditioned system of equations implying the necessity of using a direct solver for the resulting linear system of equations. We investigate a partitioned black-box coupling approach based on the idea of domain decomposition techniques. The individual problems are solved separately in an iterative manner such that we can use standard iterative solvers for the linear systems. Our approach is based on the open-source library preCICE (www.precice.org) allowing us to reuse existing solver software and simplifying the setup of new solvers that are immediately prepared for high-performance parallel computations.

Solving coupled free and porous-media flow with preCICE and DuMuX

We present recent results of coupling free and porous‐media flow applications and the development of the corresponding adapter. The main focus is on simulations based on DuMuX ( https://dumux.org/ ) which is an open‐source framework for solving flow problems, especially porous‐media flow. We present results using the partitioned approach of preCICE for different scenarios and compare it, where applicable, with monolithic simulations or exact solutions.

Contact