CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   DNS using fluent (http://www.cfd-online.com/Forums/fluent/45251-dns-using-fluent.html)

 adi July 3, 2007 13:50

DNS using fluent

Can we use Fluent for direct numerical simulation?

 Sham July 3, 2007 20:05

Re: DNS using fluent

i'd like to know too. i asked this question before but no one answered. hope someone is kind enough to shed some light here.

 PTK July 3, 2007 20:35

Re: DNS using fluent

You simply use laminar flow - ie. directly compute the turbulence scales rather than modeling them(RANS). LES is a combo.

You need a very very dense mesh and I think the unsteady solver hence incredibly long times to do it properly.

 kk July 4, 2007 23:35

Re: DNS using fluent

I can't understand why using laminar flow is using DNS method. Can laminar flow model account the turbulences in a non-laminar flow? I am also having a hard understand how laminar flow is solved in Fluent.

 PTK July 5, 2007 11:01

Re: DNS using fluent

1)DNS isn't a method it simply means you don't use a turbulence model. You need incredibly dense grids though.

2)You need to look up a book for more info. As far as I know the laminar flow(which isn't a model) simply means viscous flow and so uses the Navier Stokes equations and the invscid flow doen't consider the viscosity and uses the Euler equations.

3)Laminar flow just solves the Navier Stokes equations without using a turbulence model. By using a very very fine grid you can directly compute the turbulence scales. RANS turbulence modeling just calculates a value for the turbulence that can be used by other function in Fluent.

I would advise you to read the fundamental(first few) chapters of any CFD book. Especially Computational Methods for Fluid Dynamics by Ferzier.

Please be aware I have no DNS experience so the above answers may need to be slightly modified.

PTK

 Eugene July 9, 2007 14:12

Re: DNS using fluent

Almost DNS modelling will be under following conditions: At usage LES of model in 3d case, In unsteady setting, With very mesh sizes of a mesh of a rated net. Such that the minimum size of turbulent pulsings (Vortexes) was more mesh width. All spectrum of turbulent pulsings (Vortexes) here is considered.

 Eugene July 9, 2007 14:14

Re: DNS using fluent

Almost DNS modelling will be under following conditions: At usage LES of model in 3d case, In unsteady setting, With very mesh sizes of a mesh of a rated net. Such that the minimum size of turbulent pulsings (Vortexes) was more mesh width. All spectrum of turbulent pulsings (Vortexes) here is considered.

 All times are GMT -4. The time now is 00:39.