History of CFD
 

Please could somebody tell me where I can find some history of both fluid dynamics & evolving into computational fluid dynamics, I need it for my final year project at City University desperately.

Re: History of CFD
 

(1). CFD Forum is a good place to start. It has covered the CFD history for almost ten months now. (2). Look for a listing of CFD related books using the Forum and the resources. This will tell you when the book is published and what subjects are covered. (3). Get a listing of CFD related journals. Library is probably the best place to find the available CFD related journals. You can find out when the journal was first published and the subjects of paper covered since. (4). Do the same thing for the Fluid Dynamics books and journals. (5). Then you can write a paper on the CFD history which should include the subject, author, book, journal, abstract, and the date. (6). Since almost everyone has his (or her) own definition of CFD, it is a good idea to find out your definition also. Have a nice trip to the history of CFD !

Re: History of CFD
 

Tony, as John pointed out there are many versions of the history of CFD and many important contributors. I would like to suggest that you look into the work done at Imperial College in UK on finite volumes since their contributions were a very important part in CFD's history. A way to approach this project may be to start with the oldest papers you can find that describe early finite element approximations then describe the advances of solution methods as well as grid generation up through the present touching on important historical events like advances in computing power, application of k-e turbulence models, use of finite volumes for mass and momentum conservation, and you may also find value in including something about how various fields such as aeronautics & astronautics research have impacted CFD. You might finish your history with brief descriptions of current commerical codes and the recently released public domain codes and their specific applications. Sounds like fun! Please post a message when you're finished and let everyone know how it turned out! :)

Re: History of CFD
 

Tony,

There has been some articles on CFD detailing its history in Scientific American and Mechanical Engineering. I believe Parviz Moin wrote the Scientific American article. Also, don't forget about Richard Feyman's 1947 article (I think it was Scientific American) that basically started modern CFD as we know it.

Thanks, Tony

Anthony C. Iannetti

Re: History of CFD
 

(1). Before the man landed on the moon in 1969, I was not aware of the modern Computational Fluid Dynamics. (2).In 70's, the modern CFD started taking shape with body fitted coordinates system, two-equation turbulence models, numerical algorithms for high Reynolds numbers, 3-D flows, incompressible as well as compressible flows. (3). In 60's, the main effort was in the solution of the boundary layer equations which is no longer part of the modern CFD. Even in the early 70's, the computations were carried out on Cartesian or cylindrical non-body-fitted coordinates.(4). So, I think, the modern CFD started after the man landed on the moon, that is in 70's. ( we are mainly talking about the 3-D solution of high Reynolds number Navier-Stokes equations which requires mesh generation, turbulence modeling etc. )

Re: History of CFD
 

I took a class on CFD and the professor handed out a list of "Significant Milestones in CFD". This class was at Purdue and taught by Joe Hoffman. I will say this is not a complete list, but almost all are generally significant. Here it is:

1. Finite difference approximation of the Laplace (or Poisson) equation and solution by iterative methods [Richardson (1910), Liebman (1918), Southwell (1940) and Frankel (1950)].

2. Stability analysis of parabolic and hyperbolic PDE's [Courant, Friedrichs, and Lewy (1928)].

3. von Neumann method of stability analysis of FDE's [O'Brien, Hyman, and Kaplan (1950)].

4. Conservation law form of the governing equations of fliud dynamics [Lax (1954)].

5. Alternating Direction Implicit (ADI) methods for elliptics PDE's [Peaceman and Rachford (1955), Douglas and Rachford (1956)].

6. Lax-Wendroff method [Lax and Wendroff (1960), MacCormack (1969)].

7. Numerical solution of vortex shedding from a cylinder [Harlow and Fromm (1965)].

8. Steady state solutions from the asymptotic steady state solution of a corresponding unsteady problem [???, early 1970's].

9. Modified differential equation concept [Warming and Hyett (1974)].

10. Time linearization and approximate factorization (AFI) concepts for implicit time marching methods [Beam and Warming (1976)].

11. Flux vector splitting methods [Steger and Warming (1981)] and flux difference splitting methods [Godunov (1959), Chakravarthy and Osher (1983)].

Personally, I'd like to add some:

12. SIMPLE: a solution methodology for the coupled mass and momentum conservation equations in incompressible flow [Patankar and Spalding (1972)]. Many commercial CFD codes use this method or variations on it to this day.

13. The k-epsilon turbulence model. I can't reference it quickly, but to my knowledge it seems to be the most commonly used turbulence model for multidimensional flows.

14. Finite volume methods, specifically for unstructured grids.

15. Multigrid methods

That's all for now...

Re: History of CFD
 

John,

Your definition of 'Modern' is very much different than mine. I am defining modern as the use of electronic computers to solve the equations of fluid dynamics. This is exactly what Feynman outlined. Solving the equations of fluid dynamics, albiet simplified, were done before this period on mechanical computers. Even before this, large groups of people with slide rules were utilized as parallel people computers. Even today, since we are not doing DNS to simulate problems of real interest like airplanes, cars, etc, or since we do not have that perfect theory of turbulence, CFD really is a model. Many people could argue with you that most CFD calculations today are essentially one dimensional, because of the use of wall functions. We could argue about this forever. I see your point, but I feel that the core of CFD was build in the 1940's.

Thanks, Tony

Re: History of CFD
 

(1). I would say that most modern commercial CFD codes are based on the methods developed in 70's and 80's. (2). I would be interested in seeing a paper or a book published in 40's using the word "CFD". (3). If a paper is published in a CFD journal, then it is safe to say that it is a CFD paper. If a book has the word CFD in its title, then it is a CFD book. (4). One of the most important issue of modern CFD is the mesh independent solution issue. I don't think it was identified or even possible in early 70's. This issue was addressed in mid-70's for a very simple square cavity problem for Reynolds number up to around 400. (5). I agree with you that fluid mechanics is a much broader field than CFD. In the area where computer is used on fluid mechanics problems, CAFD ( Computer Aided Fluid Dynamics ) seems more consistent with CAD ( Computer Aided Design ) and CAE ( Computer Aided Engineering ). (6). The definition of the modern CFD can be derived from the description and the contents of commercial CFD codes being used currently.

Re: History of CFD
 

(1). None of the papers and books mentioned has the word "CFD" in the title. (2). Therefore, it is not up to us to change the classification of the paper and the book, unless they were published in the CFD journals. (3). I don't think the numerical algorithms alone can be qualified as CFD papers. If that is the case,then CFD would be used in the paper's title. (4). The development of numerical algorithms is important to CFD. But CFD is not the development of the numerical algorithms. The same is true for the geometry and mesh generation. Because the geometry and mesh generation can be used for design, structure analysis, electromagnetic problems etc., which are not related to fluid dynamics at all. (5). So, it is very important to recognize that modern CFD is an integrated technogy of computer hardware, system software, geometry and mesh generation, numerical solver algorithms, solution procedure control, turbulence modeling, and graphic presentation and animation. Wall function study alone is not CFD, but its proper implementation in the CFD computer program is part of modern CFD activities.

Re: History of CFD
 

CFD is the use of numerical methods to solve the equations of fluid dynamics. This can be done on a computer or by hand or some variation in between. In this light, the word CFD may not have evolved until the 70's or 80's (thus it wouldn't show up in any titles of books or papers), but that does not mean that the numerical solution of fluid dynamics was not done.

Consider the paper(s)... "Numerical Calculation of Time-Dependent Viscous Incompressible Flow of Fluids with Free Surface" (Harlow,Welch,1965) or "The Particle-In-Cell Method for Hydrodynamic Calculations"(Evans, Harlow, 1957)

These are both papers detailing the use of numerical methods to solve fluid dynamic problems and they show that the area was alive before man walked on the moon.

Re: History of CFD
 

I agree with you. The question was very specific and it is about CFD not the numerical solution of fluid dynamics equations. But you do have to agree with me that modern CAD is no longer a 2-D drawing program. Even if you can obtain the mesh independent solution by hand or calculator, it will be misleading to call it CFD. This does not mean that the work done by hand is not important at all.

Re: History of CFD
 

I realize I wasn't clear on a number of points, so I'd like to add some explanation.

I wasn't responding to John Chien's post specifically, just continuing the thread. The original post asked for some history of both fluid dynamics and how it lead into CFD. Most of the listed milestones deal with the solution of the equations of fluid dynamics, although the methods can be generalized to other problems as well.

Some of these methods were developed to deal with conduction heat transfer problems, but the parabolic and elliptic classification of the equations make these methods directly applicable to solving fluid dynamic equations (see "Numerical Heat Transfer and Fluid Flow" by Patankar). To paraphrase and old cliche, no field of research is an island. An advance in one field can affect others very easily.

The milestones listed are not titles to papers, just single sentence abstracts.

In response to John's requirement that the paper or journal have the acronym CFD in the title is far too restrictive. Prof. Hoffman included a list of 38 references to his class notes and only one has "Computational Fluid Dynamics" in it and that is the book by P.J. Roache, which was published in 1972. However, the paper by Harlow and Fromm (1965) is titled "Computer Experiments in Fluid Dynamics" (Scientific American, Vol. 212). Godunov's paper is titled "Finite-Difference Method for Numerical Computation of Discontinuous Solutions of the Equations of Fluid Dynamics" (1959). Even though this is a translation from Russian, it is clearly a CFD paper.

In my humble opinion, the beginning of CFD can be marked with the paper by Courant, Friedrichs and Lewy in "On the Partial Difference Equations of Mathematical Physics" (1928, translated from German). To my knowledge, this is where the Courant condition was first presented, which directly addresses the time step limit when numerically solving a time-dependent problem using the method of charateristics, which was a fundamental method of CFD for many years. The Courant condition also shows up with most other explicit solution methods.

Re: History of CFD
 

(1). The specific goal in early 70's to develop CFD was very clear, that is : " to replace the wind tunnel testing ". (2). To replace the testing, the CFD results must be better or as good as the wind tunnel results. (3). I can only say that, the original goal has not been achieved yet. (4). The basic problem areas are: a). complex geometry and mesh generation, b). turbulence modeling, and c). computer speed. (5). The numerical methods may have been around for a long time, but in reality, the reliable CFD has not been born yet. (6). The thirty years of cancer research does not mean that a cure has been found. (7). Working on CFD research does not mean that CFD has produced accurate and reliable results. (8). It is very bad to give an impression that CFD is the development of numerical methods for solving fluid dynamics problems. Let me repeat it again : The original goal of CFD development was to replace the wind tunnel testing. But it is not there yet. (not because there is a lack of numerical methods.)

Re: History of CFD
 

Find

Fornberg B. (1993) Computing steady incompressible flows past blunt bodies: A historical overview,

in Numerical Methods for Fluid Dynamics IV, MJ Baines and KW Morton (eds), Oxford Univ Press, 115-134.

Note that this is only part of the CFD history!

Re: History of CFD
 

I must disagree that nothing happened in CFD in the 60's. CFD in the 60's was dominated by the work of Frank Harlow's group at the Los Alamos National Lab. They trace their roots to Von Neumann's WWII work at the same lab.

The Marker-and-Cell (and its derivitives leading directly to the VOF free surface techniques), ICE and ALE, ancestors of the KIVA codes, computational chemistry and fluids on arbitrary grids, all started at LANL in the 60's. True, aerodynamic CFD started at NASA labs in the 70's, and the British at Imperial College certainly are direct ancestors of many of the commercial algorithms used today.

But Frank Harlow was a major contributor in the 60's. He wrote a seminal Scientific American article in that time frame as well. Also, much of Tony Hirt's ideas were developed at LANL in that time frame. If you're a younger fellow, do a name search on "Harlow" and "Hirt" in the archival literature. There are gems there!

Re: History of CFD
 

(1). I am aware of their work, in T3 group? (2). Tony Hirt used to work at LANL, and in early 70's he was working at SAIC, La Jolla. Later he left and work for himself, a company? (3). There are two points I was trying to make:a). to define the scope of modern CFD, thus open more doors to more people interested in CFD, whether it is CAD, geometry modeling, mesh generation, different numerical schemes, solvers, turbulence modeling, and computer graphics,etc. It is an integrated technology. So, we are all in one big family. b). We have to respect these researchers or pioneers( apparently you think they are.) because it is not our right to re-classify their publications into the CFD domain or any domain, unless they are happy to do so. Some people would like to be identified in the numerical mesh generation field, and others would like to be associated with second-order turbulence shear stresses model family. And I don't know what people at LANL do behind the door. I also don't know whether their department name was CFD Dept. or not. If they were not working in a CFD department, and the published papers are not titled CFD analysis..., then it is hard for us to say that they were working on CFD projects. Based on the news report recently, I can safely say that they were working on something else in 60's. They could be working on CFD, but I have no idea. (and if some day CFD becomes no longer a fashion ? what would we do? put them in DFD? Digital Fluid Dynamics? ) I am just trying to keep the Forum moving, that's all. Have a nice weekend !