- Message of Acceptance of JSME's Computational Mechanics Award for 1995
Message of Acceptance of JSME's Computational Mechanics Award for 1995
Brian E. Launder
Mr. Chairman, distinguished officers of the JSME, ladies and gentlemen:
May I first say what a very pleasant surprise it was to learn that my name had been proposed as one to receive this award. My thanks are most warmly and appreciatively extended to those responsible for this decision, both to those nominating me and those responsible for selecting me. It is a matter of great personal regret that I am unable to be present to receive this award in person as, nearly a year ago, I undertook to provide lectures for the Centre de Recherche en Calcul Appliqu・(CERCA) in Montreal at precisely the time of your meeting. It is, however, a great pleasure to know that my friend Professor Nobuhide Kasagi has agreed to accept the award on my behalf and convey this message to you.
Perhaps, for many of us in middle age, the route by which we have arrived at our present position appears to be by way of a succession of accidents or events over which we have had no control. In my case, I graduated in 1961 knowing that I wanted to do experimental research in boiling heat transfer. Unfortunately, my university, Imperial College, London, was being re-built at the time so I decided to go to MIT for graduate research thinking that it would be possible to join Professor Rohsenow's team. However, no assistantship could be offered me in his laboratory, so I ended up doing experimental research in the Gas Turbine Laboratory on boundary layer laminarization. On completing my doctorate, I returned to Imperial College as a Lecturer and plunged into several experimental research programmes. That could have remained the centre of my work had not the head of the group (Professor D. B. Spalding) and his exceptional research student (S. V. Patankar) produced, in 1967, a two-dimensional finite-volume solver for the boundary-layer equations. This solver was structured to be easy to learn and adapt to a wide range of flow circumstances; but the turbulence model employed was primitive. So two of my first research students, W. P. Jones and K. Hanjalic' set about providing something better: turbulence models based on transport equations. I consider myself very fortunate to have had such creative co-workers early in my career. Both have gone on to become professors and establish a strong international reputation in their own right.
Perhaps one other career change is worth mentioning. Having emigrated to the USA in the mid-1970's and settled in comfortably at the Davis campus of the University of California, in early 1979 my calm days kept being interrupted by the then Principal of UMIST suggesting I should return to the UK to take the Chair in Thermo-Fluids at his university. Eventually, I agreed to go; and, despite the succession of political and economic setbacks that took place in Britain in the 1980's, it is a move I have never regretted. What has principally made these last 15 years so enjoyable has been the opportunity to interact with colleagues with similar aspirations and complementary research talents. At UMIST, in particular, I have benefited from the insight and skills in numerical methods of Professor M. A. Leschziner. His talents in evolving numerical methods have enabled me to focus increasingly on problems of modelling, not to mention renewed work on experiments.
From the second half of the 1980's I have considerably benefited from and greatly value the friendship of colleagues in Japan. In the field of turbulence modelling, mention is especially made of Professors Nobuhide Kasagi, Hiroshi Kawamura, Yasutaka Nagano and Akira Yoshizawa. In experiments and applications relating to heat transfer, Professor K. Suzuki has generated a rich seam of interesting and valuable discoveries while, in the field of building aerodynamics, Professor Shuzo Murakami's work in applying CFD has always been very stimulating. Indeed, it was thanks to him that I paid my first visit to Japan in 1987 spending several weeks at his laboratory as his guest. There have been many others with whom I have had valuable and warmly remembered interactions, especially around the time of the 9th Symposium on Turbulent Shear Flows in 1993, organized so successfully by Professor Suzuki and his colleagues in Kyoto. Finally, in acknowledging my indebtedness to colleagues in Japan, I would express my warm appreciation to Professor Kasagi for his stimulating interactions with me as co-editor of the International Journal of Heat and Fluid Flow.
In broader terms, it is also appropriate to compliment the research efforts made by Japan in computational fluid dynamics. While Japan's leading position in computer hardware has certainly helped its ability to carry out CFD, it is the intelligent, imaginative use of these resources by Japanese computational mechanicists that have had such an impact in advancing our understanding of and ability to compute turbulent flows.
Perhaps before closing, a word or two should be said about where, from my personal perspective, computational efforts in turbulence modelling are likely to go over the next ten years or so. Some see closure efforts, based on the Reynolds-averaged form of the equations of motion, as being about to be superseded by large-eddy simulations. That cannot be the case! It is more than 30 years since Deardorff's first work on large-eddy simulations of channel flow and, while the problems tackled are now vastly more impressive, this is not an approach to turbulent flows that, for most problems, will be adopted over the next decade. This can be seen from the fact that, in one of the most advanced sectors of industry, gas-turbine propulsion, most industrial CFD work is still based on the mixing-length hypothesis. What the availability of LES does set, for those working with the Reynolds equations, is an upper bound on the sensible degree of sophistication of the turbulence model. In my view it is not sensible to solve more than one second-rank turbulence tensor from transport equations - and that tensor must surely be for the Reynolds stresses themselves. Thus, I would not recommend (as some do) the use of two sets of stress-transport equations (one for the large-scale, one for the fine-scale motion) or transport equations for the stress dissipation rates as well as the stresses. If one has a problem where it is necessary to use such an elaborate treatment, perhaps LES, with a good sub-grid-scale model, offers a more attractive route.
Finally, let me remark that, like turbulence itself, strategies for the modelling of turbulent flows need to remain highly diverse in character. It is as well to remember that the arrival of the automobile has not rendered the bicycle obsolete; still less has the coming of aviation eclipsed the demand for the car. Moreover, design improvements to all three modes of transport continue in parallel. I propose, Mr. Chairman, that a similar "heterogeneous" approach to the computational modelling of turbulence is also needed for the 2lst Century.