Turbulent Flow Simulation on HPC-Systems

Term

Winter 13/14

Lecturer

Dr. rer. nat. Philipp Neumann, Christoph Kowitz, Felix Schranner, Dmitrii Azarnykh

Time and Place

t.b.a.

Audience

Computational Science and Engineering, Informatics (masters), Mechanical Engineering (masters)

Tutorials

t.b.a.

Exam

immanent

Semesterwochenstunden / ECTS Credits

3SWS/ 5 credits (IN, CSE) / 4 credits (MW)

TUMonline

t.b.a.

---

Moodle

We will put all further announcements, worksheets, slides etc. on the courses moodle page

Announcements

For ALL students (also mechanical engineering): Please register in TUMonline using the informatics module number IN3400! If you encounter problems, please contact Philipp Neumann.

For CSE students: This course is expected to be accepted for the catalogues "D2 Computational Fluid Dynamics" and "E3 Parallel and Distributed Computing, High Performance Computing". A respective proposal has been sent to the CSE examination board.

Contents

The simulation of turbulent flows is a challenging problem: turbulence models are required to capture the correct physics and high-performance simulation codes are needed to exploit massively parallel systems in order to simulate relevant scenarios.

This course brings together students from computer science (IN) and mechanical engineering (MW) who work together in interdisciplinary teams to tackle this challenge:

• computer scientists are experts with respect to efficiently solving numerical algorithms on different (e.g. massively parallel) platforms
• mechanical engineers are familiar with the theory on turbulence modeling and the physics of flow problems

The student teams (2-4 persons with 1-2 persons from each discipline) work on a three-dimensional Navier-Stokes solver and

• model the data structures of the solver,
• extend the solver for distributed parallel simulations using MPI (message passing interface),
• incorporate turbulence models into the software.

Depending on the current tasks, students from IN or MW take the role of the "teachers" within the team. The work is accompanied by lectures on both

• the scientific background in computational fluid dynamics and high-performance computing as well as
• soft skill courses to enhance the interaction inside the team.

At the end of the course, each team should have a running 3-D turbulence simulation code.

Wind tunnel experiments are to be conducted at the end of the term to compare both simulation and experiment.

The teaching concept of this course Teamwork Across Disciplines: Interdisciplinarity Meets Supervised Teaching was awarded the Ernst-Otto-Fischer Lehrpreis in 2012. Further information can be found here.

The proposal of the concept can be downloaded here.

Prerequisites

Mechanical Engineering:

• Fundamentals of numerical fluid mechanics (MW0603, MW1913)
• Turbulent flows (MW0595) + prereq.

Computer Science/ CSE:

• Numerical programming (IN0019) and/ or
• Parallel programming (IN2147) and/ or

Evaluation/ Grading

The grading is accomplished similar to lab course by regular review sessions. In each review session (about 20-30 min), the team members present their work and are orally examined by the advisors.

The weighting of the different review sessions is to be announced at the beginning of the course.

Lecture Notes and Material

lecture	material

Literature

• D. Wilcox, Turbulence modeling for CFD
• P.A. Libby, Introduction to turbulence
• S.B. Pope, Turbulent flows
• M. Griebel, T. Dornseifer, T. Neunhoeffer, Numerical simulation in fluid dynamics: a practical introduction