Computational Fluid Dynamics requires coding

Editorial

Shouldn't an achievement like copyright be introduced into scientific marketing as well? Doesn't the FORTWIHR need an institution like GEMA to protect its rights? Thanks to excellent contacts with our FORTWIHR alumni, who are now working abroad (at least as the TUM President reads - i.e. outside of Bavaria), our busy attention was drawn to the website of the Institute for Scientific Computing at a (in the above sense) foreign technical university . And once again the view across the borders brings enlightenment and amusement, as it reads, among other things:

In the still young research area of ​​Scientific Computing, the technical expertise of engineers, the numerical methods of mathematicians and the modern methods and computers of computer scientists are used in an interdisciplinary manner. Many questions from scientific and technical applications can only be solved through simulation, for which mathematical methods and algorithms have to be developed. Due to the large amount of computing required, vector computers or massively parallel computers have to be used. The efficient use of such supercomputers still requires a lot of specialist knowledge that not every user can and should acquire. (True to the motto: So that we can still get third-party funding from industry tomorrow; editor's note.)

... The Institute for Scientific Computing provides this specialist knowledge and, in cooperation with users from other institutes and institutions, solves problems on its own or external parallel computers. ... The institute represents this interdisciplinary field of work and research at the intersection of mathematics, computer science and natural and engineering sciences in research and teaching.

In view of such a close reference to FORTWIHR ideas and prospectuses, we would like to hold up our hands now - after all, such formulations always require a certain amount of time despite their imperfection. On the other hand, it must be seen as a compliment if the FORTWIHR's message is heard and copied all over the world. But even in the almost verbatim passages of the above excerpt, significant deviations can be noted - well, which FORTWIHR connoisseur sees the difference? Right, the interdisciplinary commitment balanced to be done. But they will still feel the consequences of violating the parity principle. In addition, the FORTWIHR is known to move in Field of tension of the disciplines mentioned, the above institute in the Intersection. Admittedly, the area of ​​tension is not overly ingenious (we have had to listen to that several times), but a single small point as a field of activity of an entire institute now seems a bit poor. If it does exist: is the existence of such an intersection even guaranteed? Question after question. But perhaps as a TUM computer scientist you shouldn't go too far on this point problem - I vaguely remember a colloquium held in Munich in June 1989 "Computer science at the intersection of numerical mathematics, computer design, programming, algebra and logic" - mine at the time first such event as institute novice. And even then I wasn't the only one who was amazed!

But back to Germany. The three-year funding period of FORTWIHR II ends on May 31st. But don't worry, as long as there is something comparable - whether it is "FORTWIHR III", "Transfer projects within the framework of FORTWIHR" or "Actually nothing to do with FORTWIHR individual projects that but still somehow called FORTWIHR "- that is how long it will be Quartl (or transferquartl?) give.

Hans-Joachim Bungartz

Numerical calculation of the correction of hot wire measurements near the wall

Hot wires are used in experimental fluid mechanics to measure local flow velocities.

The calibration of the hot wires takes place in free currents, so that measurements close to the wall are consequently subject to measurement errors, since the proximity to the wall was not included in the calibration. Hot wires close to the wall therefore require wall corrections that differ depending on the conductivity of the wall material. The following illustration shows a greatly enlarged view of a hot wire probe. Typical hot wires have a diameter between 0.6 and 60 µm and temperatures in the range 150-300 ° C. Due to the conditions described and typical wind tunnel speeds of 5-30 m / s, the Reynolds number of the flow around a hot wire is in the range 0.1 <~ re <~ 100. Under ideal conditions, the problem of the coupling between flow, speed and heat flow is simplified to a two-dimensional flow around a heated cylinder.


Example of a hot wire probe


The main aim of a dissertation recently completed at the LSTM Erlangen was the detailed numerical investigation of the heat transfer and the laminar flow around a 2-D circular cylinder under conditions that correspond to the flow around a hot wire, especially in the case of a measurement close to the wall. In order to be able to carry out the investigation described, an expansion of the FASTEST-2D flow solver at the LSTM was necessary. Since the calculations required very large computational areas with high grid resolution in the vicinity of the cylinder, a new method of local grid refinement was developed and successfully implemented. The most important results of the work are briefly summarized below:

  • To be used as a reference for studying the influence of various factors on the outcome of NU, C.D. and St. To serve, the borderline case of a free flow with a low temperature ratio (τ = 1.003) was first analyzed, specifically for area 10-4<= re <= 200. All results are in good agreement with the available analytical and experimental data.
  • By comparison with the reference data, the influence of the temperature-dependent material properties of the fluid on the NU- and C.D.Values ​​examined. A new form of representation of these parameters was proposed with the aim of reducing the results from different values ​​of τ to a single curve. The suggestion is to use all material properties of the fluid at cylinder temperature. Unfortunately, this new form of representation only turned out to be one specific reRange as valid. Fortunately, the valid one is correct reRange in the case of the Nusselt number coincides with the most important range for hot-wire anemometry.
  • The most important aspect was the investigation of the influence of the wall distance on hot wire measurements close to the wall. For two borderline cases, namely that of an ideal heat-conducting and an insulating wall, it was possible to estimate speed corrections for hot-wire measurement data. The results for the case of an ideally thermally conductive wall showed good agreement with available experimental data. As it was found to be advantageous, a new form of speed correction has been proposed which has the advantage of being limited between 0 and 1 in the case of a highly conductive wall. In addition, it directly shows the relative magnitude of the wall influence on the measured apparent speed and shows less dependence on the temperature ratio of the hot wire.
  • The fall of an insulating wall showed unexpected results. Instead of higher apparent speeds, lower values ​​were predicted. This contradicted previous experimental results. A detailed analysis of the physical solution revealed new aspects of this complex flow problem. It was shown that the existing experimental results from measurements with "non-conductive" walls actually correspond to a combination of the two calculated limiting cases, since the thermal conductivity of such walls is always greater than that of the air.
Further, ongoing investigations in this area estimate the influence of the thermal conductivity of the walls on the speed corrections. For this purpose, the coupled problem consisting of the flow around the cylinder close to the wall and the heat conduction in the wall is solved. The publications on this topic can be requested from the LSTM Erlangen if required.
Information via: [email protected]
LSTM-Lange / Breuer / Durst

DECHEMA Prize 1996

for numerical calculations of two-phase flows

The DECHEMA Prize 1996 from the Max Buchner Research Foundation was awarded to Prof. Dr.-Ing. Martin Sommerfeld awarded.

Prof. Sommerfeld received this prestigious award for his work in the field of numerical calculations of two-phase flows. The calculations carried out are based on finite volume methods for the fluid phase and on a Lagrangian part of the program, which records the calculation of the movement properties of the particles along particle trajectories. The work has now achieved a worldwide reputation. Prof. Sommerfeld is currently working with FORTWIHR and INVENT Computing to provide a high-performance computer program for the calculation of two-phase flows. This should enable the design of new or the optimization of existing processes in the process engineering industry. Detailed considerations of heat and mass transfer processes, also in connection with chemical reactions, should be made possible by the program efficiently.

LSTM thirst

High Performance Scientific and Engineering Computing

International FORTWIHR Conference 1998

From March 16 to 18, FORTWIHR organized an international conference on technical-scientific high-performance computing. This time the host was Siemens AG in Munich, one of the most important industrial partners of FORTWIHR.

For three days, around fifty lectures gave an overview of current developments in the field of high-performance computing. The focus was on the topics

  • Development of efficient numerical methods,
  • Parallel computing,
  • Visualization,
  • Grid generation,
  • Applications from the work areas of FORTWIHR.
In addition to the scientists involved in FORTWIHR, numerous relevant working groups from Germany and abroad were also represented with a lecture. In a separate session, former FORTWIHR employees who had been appointed to professorships in recent years also presented their current research work.

More information on the WWW under the URL

FORTWIHR Conference 98

as well as in the next Quartl.
Quartl-b

Error Control and Adaptivity in Scientific Computing

International Summer School in Antalya

From August 9 to 21, professors H. Bulgak from the Research Center of Applied Mathematics at Selcuk University Konya (Turkey) and C. Zenger in Antalya are organizing an international summer school on the subject of error control and adaptivity in scientific computing. The summer school is funded by the NATO Science Committee as part of the Advanced Study Institute (ASI) program.

The summer school is a two-week interdisciplinary course for young mathematicians, engineers and natural scientists working in the field of numerical simulation. The course is intended to provide an insight into one of the central topics of scientific computing and to promote worldwide contact among scientists. The program includes the following lectures:

G. Alefeld:Computing the Range of Real Functions
A. Blokhin:A New Concept of Construction of Adaptive Calculation Models for Hyperbolic Problems
C. Brezinski:Numerical Linear Algebra, Convergence Acceleration Methods
H. Bulgak:Pseudoeigenvalues, Spectral Portrait of the Matrices and their Connection with Different Criteria of Stability
P. van Dooren:Generalized Eigenvalue Problems and Applications in Systems and Control Convergence Acceleration Methods
B. Karasoezen:Mixed and Variable Step Size Structure-Preserving Integrators for Hamiltonian and Lie-Poisson Systems
A. Quarteroni:Adaptive Domain Decomposition and Multifield Methods
R. Rannacher:Error Control in Finite Element Computations
S. Rump:Validated Error Bounds for Sparse Systems of Equations
S. Steinberg:Using Problem Solving Environments to Produce Accurate Modeling Programs and Finite Volume Methods that Compensate for Bad Grids of Equations
O. Widlund:Domain Decomposition Methods for Partial Differential Equations
C. Zenger:Error Control for Adaptive Sparse Grids

Registrations will be accepted until March 25, 1998.
More information is available on the WWW:

ASI summerschool

Quartl-b

An Optimal Solution for CFD

The cooperation of FORTWIHR with INVENT Computing and with NEC European Supercomputer Systems (cf. edition 2/1997) is bearing fruit. The company NEC has recently been advertising on its own flyer with the flow code FASTEST - including colorful LSTM pictures with trains and stirrers.

A few excerpts from the text:
The CFD code FASTEST is the result of a cooperation between INVENT Computing GmbH and FORTWIHR (Bavarian Consortium for High Performance Scientific Computing) and represents a modern high performance tool which enables the treatment of demanding flow simulation and complex geometry. Its full vectorization and parallelization enables the short solution times, which make the term High Performance Computing become productive reality. These qualities make it perfectly suited to the advanced SX-4 series of vector computers by NEC ...

FASTEST combines the advantages of highest performing CFD simulations with a set of efficient tools for grid generation, which enables the use of CAD data to construct the most complex block structured grids with highest resolution. Furthermore, FASTEST provides all necessary interfaces to post processing tools, which offer easy interpretation of the results as graphics and animation ...

In particular the 1 to 4 CPU SX-4B is an optimally dedicated application server for FASTEST, offering an unparalleled ease of use combined with minimal turn around times and high end reliability.

You can see: The simulation programs developed in FORTWIHR are not only used in industry, they are even advertised!

Quartl-b

Sparse Grid Methods

Mini symposium at DDM 11

At this year's eleventh International Conference on Domain Decomposition Methods, which will take place from July 20 to 24 in Greenwich, Hans-Joachim Bungartz (Institute for Computer Science at the Technical University of Munich) and Michael Griebel (Institute for Applied Mathematics at the University of Bonn) are jointly organizing a mini symposium on thin mesh processes.

The concept of thin grids was developed in 1990 at the Chair of Computer Science V at the Technical University of Munich and has since been promoted in a number of research projects, each with a different focus. In the SFB 342 "Tools and Methods for the Use of Parallel Computer Architectures", the use of the method in numerical quadrature as well as techniques for the efficient parallelization of thin lattice algorithms were and will be investigated. In the DFG focus "Flow simulation on high-performance computers", the combination technology, a non-adaptive and extrapolation-related variant, was the focus of interest. As part of the work at FORTWIHR, the focus was and is on the adaptive numerical solution of elliptic differential equations. Since thin mesh methods have meanwhile also found widespread nationally and internationally (e.g. at the Universities of Erlangen, Bonn and Vienna as well as at the CWI in Amsterdam) and there is also a close relationship to a number of mainly approximation theoretical approaches, the mini-symposium is intended to bring together the scientists working in this field and serve to exchange new research results.

Quartl-b

Reality and abstraction

Trans-faculty symposium at the Technical University of Munich

On March 2nd, 1998 the TU faculties for mathematics, computer science and electrical engineering and information technology together with the Collaborative Research Centers 438, 342 and 331 as well as FORWISS and FORTWIHR invited to a one-day symposium on the subject of "Reality and Abstraction".

All twelve lectures were partitioned in a strictly interdisciplinary manner and were designed by members of different faculties. The FORTWIHR was there six times, with the subjects of "Low-noise oscillator", "Massively parallel computers", "Anisotropically etched microstructures", "Simulation of semiconductor components and circuits", "Modeling and simulation of micropumps" as well "Coating and injection molding of workpieces". It goes without saying that the FORTWIHR project managers from the three host faculties - Professors Bode, Bulirsch, Hoffmann, Wachutka and Zenger - or their employees got into the Bütt.

Quartl-b

Book tip

Numerical Simulation in Fluid Dynamics:
A Practical Introduction

After the original German version published by Vieweg in 1995, the introductory work on numerical fluid mechanics by the three FORTWIHR alumni Michael Griebel (now University of Bonn), Thomas Dornseifer (now Allianz Stuttgart) and Tilman Neunhoeffer (now Tecoplan Informatik Ottobrunn) is now also in English in front.The book published by the Society for Industrial and Applied Mathematics (SIAM) in the series "SIAM Monographs on Mathematical Modeling and Computation" contains 93 black-and-white and 17 color illustrations on approx. 230 pages.

SIAM, Philadelphia: ISBN No. 0-89871-398-6

Quartl-b

News from the SFB front

December was a very good month for two Collaborative Research Centers at TUM.

The application for the continuation of the SFB 342 "Tools and Methods for the Use of Parallel Computer Architectures", which has been working very successfully since 1990, whose spokesperson, Prof.Dr.A. Bode, himself a member of FORTWIHR, was only approved by the DFG with only minor reservations. This means that parallelization can now be carried out with concentrated force into the next millennium. In addition to the chair of Prof. Bode, the Chair of Computer Science V (Prof. Zenger, Prof. Huckle, Dr. Bungartz) is involved with a sub-project.

The SFB 411 "Basics of Aerobic Biological Wastewater Treatment", located in the Faculties of Civil Engineering and Surveying as well as Chemistry, Biology and Geosciences (Spokesperson: Prof. Dr. P. Wilderer) was launched on January 1st, 1998. In keeping with the interdisciplinary collaboration, the Chair of Computer Science V (Prof. Zenger, Dr. Bungartz) is also involved with a sub-project.

Quartl-b

Vacation Academy 1998

The holiday academy of the University of Erlangen-Nuremberg and the Technical University of Munich in Sarntal (South Tyrol) takes place from September 20th to October 2nd; As in previous years, there are also (this time two) courses offered by FORTWIHR lecturers. The overall program looks like this this year:

  1. Object orientation: analysis, design, programming
    Prof. M. Broy (TUM), Prof. H.-J. Schneider (FAU)
    Guest lecturer: Prof. M. Wirsing (University of Munich)
  2. Source and channel coding for digital communication systems
    Prof. J. Hagenauer (TUM), Prof. J. Huber (FAU)
  3. Applied Physics
    Prof. F. Koch (TUM), Prof. M. Schulz (FAU)
  4. Numerical methods in construction and mechanical engineering (FEM / BEM):
    Basics and Applications

    Prof. G. Kuhn (FAU), Prof. W. Wunderlich (TUM)
    Guest lecturer: Prof. E. Rank (TUM)
  5. Catalysis in research and application
    Prof. G. Emig (FAU), Prof. W. A. ​​Herrmann (TUM)
    Guest lecturer: Prof. J. A. Lercher (University of Twente)
  6. Modeling of material properties
    Prof. K.-H. Hoffmann (TUM), Prof. G. Müller (FAU)
  7. Physical basics of semiconductor electronics
    Prof. G. Abstreiter (TUM), Prof. G. Döhler (FAU)
    Guest lecturer: Prof. D. Schmitt-Landsiedel (TUM)
  8. Numerical methods of fluid mechanics
    Prof. F. Durst (FAU), Prof. Chr. Zenger (TUM)
  9. Information processing in the company of the future
    Prof. B. Bruges (TUM), Prof. P. Mertens (FAU)
Registration forms and more information are available here:

Ferienakademie-sz

Quartl-sz

FORTWIHR Intern

  • Since January is Murat Sabanca from the Institute for Mathematics of the University of Ankara at the LSTM-Erlangen as a visiting scientist. As part of his dissertation, he will primarily deal with the further development of numerical methods for hypsonic flows. He is one of the first particularly qualified students to apply for a Ph.D. as part of the summer academies in Turkey and Bosnia. Studies in Erlangen could be won.
  • Priv.-Doz. Dr. Bernd Kugelmann (Chair for Higher Mathematics and Numerical Mathematics, TU Munich) accepted a call to the Chair for Numerical Mathematics and Optimization at the Ernst-Moritz-Arndt University of Greifswald on 1.1.1998.

FORTWIHR lectures

  • Talked about "Virtual Worlds - Reality from the Computer" Prof. Dr. Dr.h.c. Roland Bulirsch on March 11, 1998 in the hall of honor of the Deutsches Museum in Munich as part of the lecture series "Science for Everyone".

FORTWIHR guests

in Munich:

  • 15.12.97, Dipl. Math. M. Bollhöfer (TU Chemnitz): An algebraic domain decomposition concept for the parallel solution of large, sparsely populated systems.
  • 18.12.97, Prof. Dr. Th. Ludwig (Ludwigs-Maximilians-Universität München): Methods for the realization of interoperable and universal tools for parallel and distributed programming.
  • 19.12.97, Prof. Dr. K. Kunisch (University of Graz): Suboptimal control for equations of fluid mechanics.
  • 12.1.98, Prof. Dr. B. Meinerzhagen (University of Bremen): Hierarchical simulation of Si and Si-Ge components.
  • 19.1.98, Dr. H. Landes (University of Linz): Efficient numerical simulation of multiple coupled field problems.
  • 20.1.98, Dr. M. Roos (Technikum Winterthur): Numerical modeling of electro-thermal-mechanical systems.
  • 26.1.98, Dr. Th. Scheiter (Siemens AG Munich): Highly miniaturized pressure sensor system for medical applications.
  • 2.2.98, Dr. U. Czarnetzki (University of Essen): Measurement of electrical field distributions in plasma boundary layers.
  • 3.2.98, Dr. F. Kupka (Institute for Astronomy, University of Vienna): Spectral methods on thin grids for the solution of partial differential equations.
  • 9.2.98, Dr. P. Bachmann (Philips GmbH): Diamond CVD / CVD Diamond.
  • 10.2.98, Dr. M. Riedmiller (University of Karlsruhe): Neural dynamic programming for control engineering applications.
  • 16.2.98, Dipl. Math. C. Teitzel (University of Erlangen): Particle tracing on thin grids.
  • 5.3.98, Dr. F. Jarre (University of Würzburg): Recent developments in internal point methods for linear programs.

in Augsburg:

  • 13.1.98, Dr. B. Mohammdi (University of Montpellier): A Unified Formulation for Shape Optimization and Flow control.
  • 10.2.98, Dr. K. Mauthe (University of Stuttgart): Topology optimization in structural mechanics.

Please note:

  • The Chair of Fluid Mechanics in Erlangen also runs the following short courses in '98:
    March 9-13:Introduction to fluid mechanics
    March 23-26:Measurements in turbulent flows using hot wire and laser Doppler anemometry
    April 20-23:NUMET'98 - Numerical methods for the calculation of flow and heat transfer problems, introduction to numerical fluid mechanics and technical-scientific high-performance computing in this area.
  • In the last edition of the Quartls announced
    "1st Munich Science Weekend: Knowledge creates fun"
    is unfortunately not feasible for the planned period (March 27-29, 1998) for financial and organizational reasons, according to the scientific editors of Bayerischer Rundfunk and the other organizers. But postponed is not canceled - the new date is a weekend in the first half of 1999.

By the way ...

  • Prof. Roland Bulirsch, Full Professor of Higher Mathematics and Numerical Mathematics at TUM, was accepted into the science class of the Bavarian Maximilian Order. He is the 12th living member of TUM since the order was re-established in 1980.
    Furthermore, Bulirsch was awarded the Medal of Merit of the Union of Czech Mathematicians and Physicists "in recognition of outstanding achievements in the field of applied mathematics, especially the mathematical aspects of high technology".
    He was also elected secretary of the mathematics and science class of the Bavarian Academy of Sciences.

    (from TUM announcements 3-97 / 98)


The subtitle "I'm ready", originally intended for our full-page ad premiere, unfortunately fell victim to censorship. In order to avoid a costly counter-notification, the second attempt "weak as a bottle is empty" was also discarded. In the third attempt, a compromise formula could be found together with the - of course unnamed - client.
Quartl-b

Ch. Halfar, April 30-1998