CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   LES vs RANS modelling (http://www.cfd-online.com/Forums/main/13055-les-vs-rans-modelling.html)

Valerio February 27, 2007 12:13

LES vs RANS modelling
 
I have a very important problem. I have to simulate the flow field inside Ranque-Hilsch Vortex Tube (RHVT). I saw that RANS models with RSM or AMS closures are used. LES never used in these problems. How do I choise RANS or LES in a general problem? (without tests and other tests over previous results)

chw March 1, 2007 05:43

Re: LES vs RANS modelling
 
RANS gives you averaged values, whereas LES can resolve vortex structures. In terms of accuracy LES is better than RANS, but not as good as DNS. It also takes longer to calculate a LES solution compared to RANS.

Paolo Lampitella March 1, 2007 09:29

Re: LES vs RANS modelling
 
Probably you have to think about the effort required by a LES simulation compared to that of a RANS one.

First of all, it depends of if you're making you're own code or if you will utilize a commercial CFD code. Second, the time, memory and computing power available. And last, but not the least, the kind of information required from the analysis.

Obviously, just from a quality point of view, the LES is usually best if a good subgrid model is choosed; for example, if you're using Fluent you should use the dynamic approach. In fact the RSM approach for the RANS eqns has so much terms to be modelled that, even if the model is of 2nd order, it is very empirically based. Moreover it could be as much time consuming as LES which, if adeguately set up, is much more "real".

For example, if you're using Fluent, and probably you need just some global quantities about a simulation that is statistically steady, and you're using a one/two processor workstation with not a lot of memory available then choose RANS.

On the other side, if you can spend time with initial and boundary condition definition, simulation and subsequent time and/or space averages, if you have memory and computing power, a good subgrid model (dynamic procedure and similar or some structural models) and the necessity to have an unsteady simulation, then LES is the right choice.

In general, if you are not already a LES user, that is conscious of the requirements and capabilities and able to handle the various difficulties hidden in this kind of simulation, choose RANS.


Valerio March 1, 2007 10:35

Re: LES vs RANS modelling
 
Ti ringrazio per le informazioni!

Paolo Lampitella March 2, 2007 04:10

Re: LES vs RANS modelling
 
Figurati!

Valerio March 2, 2007 05:01

Re: LES vs RANS modelling
 
L'Hirsch (Numerical Computation of Internal and External Flows) lo conosci? Che opinione ne hai? Lo vedo su molte bibliografie di articoli che scrivono di flussi subsonici comprimibili o transonici! Spero possa essere un adeguato testo di riferimento!

Paolo Lampitella March 5, 2007 05:05

Re: LES vs RANS modelling
 
Non saprei dirti precisamente perchŔ non mi Ŕ mai capitato tra le mani per˛, se pu˛ servire, guardando l'indice ed alcune pagine iniziali del 2░ volume da Amazon.com, mi sembra abbastanza approfondito su alcuni argomenti quali l'analisi di stabilit├ (alla quale dedica diversi capitoli) e mi sembra avere una simbologia non troppo complicata. Tuttavia mi pare sia molto pi¨ sbilanciato sui flussi non viscosi che sulle Navier-Stokes alle quali dedica solo l'ultimo capitolo del secondo volume, tra l'altro con molto poco riguardo alla modellistica della turbolenza. Considerando il costo per volume (210$ x 2 volumi = 420$!) e l'indice, l'impressione che ho avuto Ŕ quella del reference book. Al momento non mi sbilancerei di pi¨ perchŔ sono uno che compra spesso libri e non avendolo ancora comprato per me, pur avendone il desiderio, non posso certo consigliartelo. Probabilmente non Ŕ il libro da cui iniziare ma Ŕ forse il necessario elemento di una buona biblioteca. In ultimo, se decidessi di acquistarlo, dovresti considerare che a luglio uscir├ la 2a edizione dopo quasi 20 anni e pare che sia molto pi¨ orientata all'insegnamento rispetto alla prima e che ci sar├ forse anche un CD d'accompagnamento; avresti quindi da scegliere tra una vecchia edizione che forse scender├ di prezzo o la pi¨ aggiornata edizione di un libro universalmente adottato in ambito tecnico. Spero di esserti stato utile.

c120613 October 22, 2011 14:25

Hi Valerio,

I'm italian too. I can tell you that if you go on the website of Politecnico di Bari (where I was graduated), there are interesting video if you go to gasdinamica > turbolenza from the following link.

http://climeg.poliba.it/moodle/cours...page=20&page=0

Fron this video you can appreciate the difference between the RANS, LES, and DNS. The closest one to the physic is DNS, but you would require a grid refinement which needs to reach the kolmogorov scale.

Cheers

Domenico


All times are GMT -4. The time now is 08:42.