100 Volumes of ‘Notes on Numerical Fluid Mechanics’

Numerical fluid mechanics and aerodynamics, or if one likes, computational fluid dynamics, evolved since the mid 1960s, in a time span a little longer than a human generation, from a mere toy into a powerful and accepted tool of science and engineering.

The book series “Notes on Numerical Fluid Mechanics and Multidisciplinary Design” is portrayed. It originally was conceived as publication organ of the GAMM-Committee for Numerical Methods in Fluid Mechanics, but soon its scope was extended. In three sections it is sketched how the series came into being, how its aim developed, and how it evolved from the first to the present volume. The general editors and the co-editors of the series are listed and their duties are outlined. Finally acknowledgements are expressed to the publishing houses of the series, first Vieweg, then the Springer-Verlag, and especially to the persons directly involved in the publication of the series.

The NNFM series originated as publication organ of the GAMMCommittee for Numerical Methods in Fluid Mechanics. This committee was founded in (West-) Germany in 1974 and existed until 1992. Its development and the main activities - the organization of GAMM-Workshops and GAMMConferences on Numerical Methods in Fluid Mechanics - are sketched in this contribution.

This article reviews some of the work carried out on numerical methods in fluid mechanics in Germany during the past four decades: Early investigations in the sixties dealt with the extension of already existing solutions for two-dimensional flow problems to those of three-dimensional flows. In the following decades international short courses and conferences were established, and cooperation between scientists of various research centers and universities was successfully initiated and built up. The German Research Foundation generously sponsored these activities in two priority programs and in a third cooperative research program co-sponsored by the French Centre National Recherche et Scientific. In the nineties the German Science Council proposed a recommendation to the Federal Government to establish high-performance computing in Germany on an internationally competitive basis, which finally resulted in the foundation of the German Gauss Center for Supercomputing.

A short historical background about the usage of digital computers in Germany since 1956 is presented. Soon after the beginning, the demand on computing capacity increased, so that computing centers were founded at universities, research establishments and in industry. Later on the foundation of federal high-performance supercomputing centers was started supported by the government. The development of these centers is described as well as the access to them by the costumers. In addition, a high-performance computing supply pyramid was created with different levels of computing power. On this basis Germany started to participate in the construction of a European high performance supercomputing infrastructure, that is described including the goals and the proposed tasks.

Today’s computational methods are built upon physical and numerical models. Thus it is important to have an appreciation of the reasoning and thought processes that established our current understanding of the mechanics of fluids, all put in place before the age of numerical solutions. A brief sketch is given of the evolution of the ideas that led to the formulation of the equations governing fluid flow, the problems to which the equations were applied, and the efforts to solve them before computers were available. After the historical origins of the fluid-flow models are in place, the last section traces the transition undergone during the 20th Century, starting with analytical means to solve the mathematical problems that successively evolved into numerical approaches to solving them, thus leading up to the present time of the computational era.

CFD played a critical role as design tool in the 1970-80s in French laboratories and aeronautical industries with the rapid progress of computing facilities. Three short contributions by senior scientists and technologists describe how CFD developed rapidly in the French scientific and industrial community and was recognized as a mature tool in aerospace engineering for the aerodynamic design of civil and military aircraft and space vehicles like the European HERMES Space plane.

Reviewed are major developments of discrete numerical solution methods of aerodynamics at the German aerospace industry from the 1970s to the mid 1990s. Then the national project MEGAFLOW was initiated. Considered are potential equation codes, with panel and potential equation methods, Euler codes, boundary-layer methods, and Navier-Stokes codes.

Valuable results have been achieved over a wide variety of CFD fields by the Italian community during the last decades. Here attention is focused on a specific field: the numerical treatment of discontinuities, especially shock waves, in hyperbolic gasdynamic problems. Italian contributions to the explicit fitting and to the numerical capturing of discontinuities are reviewed in the following.

We glance back at the history of CFD in Asia, especially in Japan. It is quite clear that contributions from Asian countries have increased tremendously in the last 20 to 30 years of CFD research. Advanced supercomputers have been timely developed by Japanese industries and a good environment of CFD research using High Performance Computers has been supplied to the researchers especially in Japan. A number of CFD researchers either learned CFD in U. S. or in Japan and came back to their own countries in Asia. Such incidence obviously supports the CFD development in Asia. This paper describes the fact sheet to support the development and some features of the history of CFD development in Asia, especially in Japan.

The paper presents the review of the development of Computational Fluid Dynamics and Aerodynamics in the time frame 1960-2007 in the USSR and, later, in Russia. The organization of scientific investigations in the given fields is shown. Reviewed are the theory of difference schemes, splitting methods, irregular grids, and the particle-in-cell method. Finally, notes on software packages and computing systems are presented.

This chapter briefly surveys the CFD developments in the Nordic countries. The focus chosen is on the major current tools in use with just sketchy indication of the origins of these tools and how they evolved from earlier legacy codes. The developments in Sweden surveyed here are selected from four domains: fundamental studies of turbulence and transition, ship hydrodynamics, aeronautical CFD and numerical weather prediction. The developments in Norway begin with a historical survey of the formulation of models and then lead into present-day contributors of computational methods and codes. The developments in Denmark focus mainly on wind turbine aerodynamics and how the Danes have successfully developed the EllipSys3D code and became leaders in the wind energy business. In Finland at Helsinki University of Technology researchers have produced the FINFLO code and applied it to computational aerodynamics, ship flows and to the development of new turbulence models. Work at the Finnish Meteorological Institute is also mentioned.

The UK has been responsible for many seminal contributions to the general field of computational fluid mechanics and heat and mass transfer. Attempting to review all of them would be an impossible task within the space constraints imposed on us. Instead, we restrict ourselves to two particularly prominent areas, in which we can claim to have specialist knowledge, namely methods for simulating problems involving complex geometries and methods for the modelling of turbulence within the framework of the Reynolds-averaged Navier Stokes equations.

CFD in North America greatly increased its utility in the 1970’s with the development, locally and abroad, of high-resolution finite-volume methods. These methods permeated aerodynamics in the 1980’s, when Euler and Navier-Stokes models became the standard of industry; since then, aerospace engineering has remained at the forefront of CFD development. During the 1990’s, research emphasis shifted from fundamental discretizations to practical matters such as generating and using unstructured/adaptive grids, and high-performance computing. At the start of the 21st century, increasingly complex applications are driving a quest for compact higher-order algorithms; these borrow greatly from finite-element methodology.

This contribution gives an overview of methods used for flow simulation in the European aeronautical industry, where they are now widely accepted as analysis and design tools. However, in contrast to other industries, they are usually not provided by commercial software vendors but are developed, supplied, and maintained by national European Research Establishments. The status of the codes developed by the European aeronautical research centers ARA, CIRA, DLR, FOI, NLR, ONERA, and a code developed by Dassault is outlined, and some results of their application are highlighted.

Computational Fluid Dynamics (CFD) has become a vital tool for Aerodynamic Development. It has enabled sophisticated design optimization and comprehensive aerodynamic analysis of aircraft, and thus provides an effective means to cope with complex product requirements. Major applications demonstrate the essence and value of high fidelity CFD in a wide field. The outlook to future needs shows a high potential of CFD to lead to a complete change of paradigm of the work of the Aerodynamic Engineer which is strongly supported by upcoming development projects on simulation technology.

In the mid 1980s France launched an industrial programme for the development of a winged space vehicle called HERMES. Approximately three years later HERMES became an European Program under the supervision of the European Space Agency ESA. This was the focal point for the begin of activities in the frame of numerical aerothermodynamics in particular in space industry. The companies Dassault Aviation, Aerospatiale, Saab, Dornier and MBB were partners in this project and have pushed forward in a special research program the physical modelling and the numerical methods for solving the governing equations of aerothermodynamics. After the run-down of the HERMES project somewhere in 1993 several smaller international and national activities were undertaken in order to further extend and harmonize the level of competence in this aerothermodynamics. Some of the projects and programs were the Crew Return Vehicle (CRV), the Manned Space Transportation Programme (MSTP), the X-38 project, the Technology for Future Space Transportation Systems (TETRA) and the PHOENIX demonstrator. This report addresses some details of the work in the above mentioned projects.

In order to understand the vortical flow around delta wings especially with rounded leading edges and to collect new flow field data for comparison with numerical results, the Second International Vortex Flow Experiment (VFE-2) has been carried out in 2003 to 2008 within the framework of an RTO Task Group. The tested configuration was a flat plate 65° swept delta wing with interchangeable sharp and rounded leading edges. Five different models were tested in various wind tunnels worldwide. The program of work and some of the main experimental and numerical results are presented in this paper, and an outlook concerning future investigations on this configuration is given.

Stimulated by the developments in computer technology, numerical algorithms, and physical modeling large-eddy simulations have become more and more mature over the last couple of years such that they can be used to analyze even highly complex flow fields. In the following, a number of industrially related flow problems ranging from open channel flows to combustion chamber flows will be briefly discussed on the one hand, to underline the maturity of the large-eddy simulation approach and on the other hand, to get more and more engineers and scientists interested in the concept of large-eddy simulation.

Issues of multidisciplinary design are considered in view of the aerodynamical and structural design of the airframe, i. e. fuselage, wing, and tail unit. Background problems, like Cayley’s design paradigm are discussed, as well as ideal-typical airframe definition and development phases, and the industrial challenges which numerical multidisciplinary design and optimization (MDSO) poses. Finally the state of the art of MDSO methods is illustrated.

One of the new challenges in aeronautics is combining and accounting for multiple disciplines while considering

uncertainties

or variability in the design parameters or operating conditions. This paper describes a methodology for robust multidisciplinary design optimisation when there is uncertainty in the operating conditions. The methodology, which is based on canonical evolution algorithms, is enhanced by its coupling with an uncertainty analysis technique. The paper illustrates the use of this methodology on two practical test cases related to Unmanned Aerial Systems (UAS). These are the ideal candidates due to the multi-physics involved and the variability of missions to be performed. Results obtained from the optimisation show that the method is effective to find useful Pareto non-dominated solutions and demonstrate the use of robust design techniques.

With the growing interest in environmental issues, the automotive industry is required to implement a range of measures such as increasing fuel efficiency or decreasing pollutants from exhaust gases. It is no doubt that CFD is an encouraging technology to develop an innovative idea by providing valuable data which conventional experimental methods can not measure. Compared with other industries, the automotive industry has been taking the initiative in introducing Computer Aided Engineering (CAE) at various stages of manufacturing. Thus, we can look into the state of the art of engineering CFD by reviewing its applications in automotive engineering.

In the developed countries the installation of new hydro electric power stations is nowadays very difficult and in many cases impossible. Here the upgrading of old power stations through replacement of critical components is the right measure to considerably increase the production of electricity by use of renewable resources. The development of such improved and more powerful components, most of all the turbine runner, is today possible based on numerical flow simulation and other modern CAE-tools. More than 50 % of old power plant installations have operational problems such as cavitation erosion, vibration of the structure due to vortices at off-design operation or severe noise. As an example the numerical engineering is described, that was carried out in order to upgrade a low head small hydro electric power station in Germany to increase power production by 30 % and to avoid cavitation erosion in the turbine runner. The new components have been installed and successfully put to operation. Now the turbine is performing well and is running surprisingly smooth.

A brief overview of the state of the art of computing blast loads on civil engineering structures is given. The general problem setting, requirements, main physical phenomena and timescales, as well as suitable numerical methods are described. Several examples show the power of blast loads calculations for civil engineering structures.

This contribution gives a short overview over modern numerical combustion modelling. Numerical simulation of combustion is a multi-scale problem, because the specific issues of fluid mechanics and chemical reaction systems accumulate. There exist a large variety of combustion models for different flame types, which are more or less universal. For some turbulent reacting flows, existing methodologies are acceptably accurate, and have justifiable computational cost. Depending on the expected answers of numerical simulation, substantial advances are required and have to be worked out.

In this contribution the authors address recent advances of modelling and simulating flow problems related to Environmental and Civil Engineering using Lattice-Boltzmann methods (LBM) and present results documenting the potential of this kinetic approach. After a short introduction to theoretical aspects of the method, we address extensions of the basic LB ansatz to model turbulent, thermal and multiphase flows, free surface flows and bidirectional Fluid-Structure-Interaction. All simulations were done with the LBM research prototype software Virtual Fluids [16], a transient 2D/3Dcode offering adaptive hierarchical Cartesian grid refinement, massive parallelization and multi-physics capabilities. The simulation of a simplified debris flow problem is based on the coupling of the flow solver with an external physics engine.

Process engineering focuses on the design, operation and maintenance of chemical and material manufacturing processes. As in other disciplines, CFD has attained a certain maturity in order to analyze and optimize these processes. Besides convective, turbulent and diffusive transport of mass, momentum and energy, CFD in process engineering has to account for several other complexities such as multi phase fluids including phase changes, the kinetics of chemical and biological conversion, unknown material and thermodynamic properties, among others. The paper presents a selection of applications, where classical CFD techniques as well as new developments such as the lattice-Boltzmann technique are utilized.

The progress during the past forty years of computational electromagnetics in the time domain is summarized. Contributions from the computational fluid dynamic community to this scientific discipline are highlighted. The impact of characteristics-based formulations, high-resolution algorithm development, and concurrent computational techniques has alleviated two fundamental limitations in computational electromagnetics. This knowledge sharing has opened new avenues for basic research and practical applications for electrical engineering and interdisciplinary computational physics.

Magnetic plasma confinement research poses multifaceted requirements for computational modelling. The determination of plasma equilibria, the prediction of their global stability, the physics of plasma heating, the estimation of the energy losses out of the plasma and the interaction of the energetic plasma with the walls require all support by modelling, using distinct approaches. In particular, the quantitative analysis of turbulent energy transport was for a long time exclusively based on semi-empirical approaches. In the last decade, however, ab-initio plasma models have become progressively more realistic. The contribution reports highlights and trends of this development since the de-classification of fusion research, and describes the components of a numerical tokamak, expected to become, concomitantly with the burning plasma experiment ITER, the main research tool of the fusion

Recent progress in geophysical modelling of global plate tectonic, mantle convection and seismic wave propagation problems is reviewed, while paying particular attention to novel adjoint methods for the efficient inversion of seismic and tectonic data. Observed is that the continuing growth in high performance and cluster computing promises the crossing of long standing barriers in the simulation of first-order geophysical phenomena.

This paper describes progress in computational plasmadynamics applied to the prediction of “space weather” - the study of how conditions on the sun lead to transients in the solar wind, which in turn affect Earth’s magnetosphere, ionosphere and thermosphere. The progress is based on advances in algorithms, parallel computing, and a software framework that couples the multi-physics and multi-scale modules necessary to model this challenging and important problem.

The matter of astrophysical objects (

e. g.

, a star or a galaxy) can often be approximated as a gas or fluid,

i. e.

, the equations of fluid dynamics are adequate to describe the astrophysical phenomena. Hereafter, for simplification, the word fluid will be used as a synonym for both fluid and gas. Because most astrophysical conditions are inaccessible in the laboratory, and as astrophysical fluid motion may occur on time scales long compared to the life span of humans or deep inside astrophysical objects, numerical simulation is the only means to study such fluid motion. In this respect numerical simulations play a more important role in astrophysics then in most other branches of physics: astronomers are passive ‘observers’ of what Nature decides to show them. The study of astrophysical fluid flows is further complicated by the effects of self-gravity, which must be considered in many astrophysical flow problems, by the enormous range of length scales and time scales to be covered in the simulations, and by a variety of other physical effects which must be taken into account frequently. The latter include radiation transport (of photons or neutrinos), heat conduction, radiative cooling, ionization and recombination of atoms, magnetic fields, energy generation by thermonuclear reactions, flow velocities near the speed of light, and strong gravitational fields.

In this article, we review the development of multigrid methods for partial differential equations over the last 30 years, illuminating, in particular, the software question. With respect to industrial software development, we will distinguish “optimal” multigrid, multigrid “acceleration” and “robust” multigrid. Surprisingly, not geometric multigrid but algebraic multigrid (AMG) finally brought the breakthrough. With the software package SAMG, which is based on block-type AMG, systems of partial differential equations can be treated efficiently also. Finally, we outline how SAMG is used for industrial applications.

The role of computer science in application fields such as computational fluid dynamics has been underestimated for a long time. However, with a growing complexity of application scenarios as well as numerical algorithms and hardware architectures, the need for sophisticated methods from computer science becomes more and more obvious. Just think of the visualization of the in general very large data sets resulting from numerical simulations, parallelization and load balancing in particular in combination with the upcoming multicore architectures and petascale computers, code verification and software engineering for large and highly complex software codes, and, of course, the efficient implementation of classical numerical algorithms which are typically data-intensive and, therewith put a big challenge on data storage and data access concepts. We focus on the last point in this contribution and show two examples in more detail, where computer science contributes to more efficient simulation environments: in the memory management for adaptive space-tree grids and, second, in the field of partitioned fluid-structure interactions.

This paper traces the development of Computational Fluid Dynamics (CFD), from its beginnings in the second half of the 20th Century to the present time with special emphasis on the evolution of the market for commercial codes. A discussion is provided on the reasons behind the current ‘success’ of commercial CFD packages and the present issues concerning the everyday use of CFD in Industry are examined. An attempt is made at providing some perspectives for its future deployment.

High Performance Computing (HPC) has undergone major changes in recent years [1]. A rapid change in hardware technology with an increase in performance has made this technology interesting for a bigger market than before. These changes have also had a massive impact on the usage of HPC in industry. Changes in architecture have had an impact on how industry uses HPC and for what types of applications. The dramatic increase in performance with a drop in prizes has changed the role of HPC as a tool in industry. The High Performance Computing Center Stuttgart (HLRS) has collaborated with various industrial companies in a private public partnership for high performance computing over the last years. This contribution describes the collaboration and the changes it had to undergo in order to keep track with the changing requirements. The paper further comes up with the most important lessons learned from such a long term collaboration with industry in the field of high performance computing.

Computers were conceived as a replacement for the mechanical solution of engineering applications using simple calculators. The basic idea of Konrad Zuse and John von Neumann was to automate this process by introducing programmability, i.e. the description of the sequence of execution of individual steps of a longer algorithm including the possibility of resultdependent branches and loops. In this way, a universal calculating tool of enormous power was invented. Today, the solution of engineering applications and the field of numerical simulation are the necessary prerequisites for advances in science, products and services. Computers are even more often used for non-numerical commercial and consumer applications today, which shows the universal applicability of the concept. In the following, this article concentrates on the application of computers for numerical simulation.

In this contribution, the past technological progress in high performance computer systems is looked back from an architectural view point. Then, major technological challenges and emerging technologies in hardware and software for Petaflops systems are discussed. Those future technologies include device technologies of CPUs, interconnection technologies in hardware, and technical challenges for application development. Finally, two typical projects of Peta-scale computing systems in the US and Japan are introduced to.