I am a novice in CFD field. I wonder if the turbulence modeling such as ASM will be replaced almost completely by LES or DNS in pratical engieering in several decades. I also want to know what is the hotspot of CFD research.

Turbulence modelling is an active field for research yet (many papers published during the last year concern turbulence effects and transition modelling), but, in my opinion, combustion modelling and aeroelasticity arouse greater interest in active research due to the fact they are almost unexplored islands.

I have a question along similar lines. At present people are conducting LES and DNS simulations, with the DNS being only on very simple geometries. Computing power is still a restricting factor in these areas and will be for the forseable future. Does anyone have an estimate on how long it will take PC's to progress and develop to the level that LES and also DNS simulations can be undertaken on Industiral projects within the associated time scales that are required in industry ? ie when will say LES be the standard option in the Turbulence modelling form of CFD software instead of say KE models.

Also, for DNS simulations and research, what are the main areas of study, apart from experimental comparisons are there other issues that need to be ironed out such as those of wall functions for LES ?

Bob

Wasn't it Spalart who predicted that it would take at least another 40 years or so before LES would be practical for "real world" problems, in his "Comments on the feasibility of LES for Wings and on a Hybrid RANS/LES Approach"? I know he said that somewhere...

But even if LES is feasible, that doesn't necessarily mean it is desirable. The time scales required industry change, usually not for the better for the engineers, and LES will always be expensive compared to RANS models. Especially if improved turbulence models (better than KE is easy ;-) can be developed in the next few years/decades, then I would be very surprised that LES would ever be the default option for your turbulence model.

Steven

At least one reference is:

Spalart, 2000, "Strategies for Turbulence Modelling and Simulations", Int. J. of Heat and Fluid Flow, v. 21, pp. 252-263.

He estimates that LES will be ready in a research (not industrial) environment by 2040, and DNS by 2080. He takes the position that hybrid LES/RANS methods are the way to go for the foreseeable future. But as you say, a RANS model that gets the problem 80% right in a day might be preferable to an LES (or hybrid) model that gets it 95% right in a month. In a few decades, maybe RANS in 20 minutes will look more attractive than LES in a day.

Keith

LES is implemented in almost any commercial code. But who uses LES on a day to day basis. I suppose that almost nobody uses this model extensively.

The hybrid approach (RANS near of walls (i.e.SST), LES elsewhere) seems to be promising. It is already implemented in some commercial codes. The switch between the two models is made grid size dependent together with a smart blending function. So one uses the existing Know-How of near wall turbulence modeling.

I came across this great quote the other day, which made me smile:

Dean Chapman Nasa 1977 in the John D Anderson book : the Fundamentals of Fluid Mechanics.

A numerical simulation of the flow over an airfoil using the Reynolds averaged Nav.-Stokes eqn's can be conducted on today's super computers in less than half an hour for less than $1000 cost in computer time. If just one such simulation had been attempted 20 years ago on computers of that time and with algorithms then known, the cost in computer time would have amounted to roughly $10 million, and the results for that single flow would not be available until 10 years from now.

Thank you for your responses.Thanks very much! I will be a postgraduate soon,but I do not know which field I should choose as my research interest. I am in China .I do not think the research level of my lab is high.Probably it can not provide opportunity for me to do research on LES and DNS.So I post the problem about the prospective about turbulence modeling research.I do not know if CFD is a good choice for me. Could you give me more advice for my choice? Thanks very much !

Hi The DNS and LES are only the approaches for Turbulence modeling. The accuarancy of these approaches are depend on the given modeling problem. But I must think we must change this phyloshopy to describe the concept of Turbulence.That is a very hard work and anybody can create the concept for Turbulence by using variuus coordinate systems.

Thanks

K.K.J.Ranga Dinesh