CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   doubt about DNS and LES of combustion (http://www.cfd-online.com/Forums/main/6791-doubt-about-dns-les-combustion.html)

chinthakindi November 6, 2003 11:54

doubt about DNS and LES of combustion
 
Hi CFD friends , I have one small doubt , regarding the DNS and LES of combustion. I have one code which is giving the good results for LES of diffusion flame(round jet) with Smagornsky model,the results are more accurate than experimental results. I removed smagorensky model from my code and increased the grid ressolution, in this case also i got good results more or less like LES results. can i call this second results as DNS of the flame or not, if it is not why? I am sincearely requesting you to all to clarify my doubt.

Thankyou in advance

Andy November 6, 2003 14:05

Re: doubt about DNS and LES of combustion
 
You cannot call this DNS of a flame, by removing the sub-grid terms you are solving the DNS equations set, but not at the correct accuracy spatially, plus many other things. The only reason that your code even continued is that more numerical diffusion is beginning added by your code than is actually in the sub-grid terms, your code is overly diffuse. There are also other difference regarding the DNS and LES, things like the way you work out viscosity in DNS is not the why it is done in your LES code. You are thinking on the right tracks LES should tend to DNS in the limit of a small enough grid size, hence people can do a priori LES with their DNS results by filtering them, but what you are doing is not right. I would also be careful with statements that suggest that your LES is more accurate than an experiment, especially when it is just standard smag.

At best you can call your results implicit LES, and worst case they are not correctly respecting the physics. The sub-grid terms are there as an energy drain and are required, fine, you may achieve this with numerical diffusion, but they should be there, since if you reduce your numerical diffusion by going to higher order, then they will be required again.

Andy

chinthakindi November 7, 2003 05:23

Re: doubt about DNS and LES of combustion
 
Hi Andy, Thankyou very much for your comments, According to you, if I decrease my numerical diffusion by high order, then i need again SGS model. suppose now again if remove SGS model,and if i get good results , can i call DNS at this time or not, if not why? finally when exactly we can call our results are DNS. what i want to know is, i want to DNS with utilising this code which is giving good results for LES, can u recommend the way to achieve my goal.

thankyou

jdc November 10, 2003 06:05

Re: doubt about DNS and LES of combustion
 
Hi,

I think Andy's comment were stating differencies between LES and DNS in terms of aerodynamic. To asses the possibility of your LES code, homogeneous isotropic turbulence can help. The behavior to expect are the following:

When using a LES grid (not to the Kolmogorov scale), run your code with and without SGS:

-> your code is too dissipative for DNS if (1) the turbulence spectra does not show an accumulation of energy at the smallest scale and if (2) it does blow up when SGS is off.

-> plot the time-evolution of the turbulent energy for both simulation. There should be some major differencies.

Apart from that, DNS and LES of combustion are also very different in terms of modeling. Combustion models for LES keep in mind that combustion usually occurs at sub-grid scale, so that PDF models are common place. For DNS, combustion occurs at scales larger than the grid scale. So, the combustion process can (need to?) be more accurately described. A litterature comparison between LES and DNS of combustion will clarify my statement.

Hoppe this may help. Julien

JAY ADHVARYU August 17, 2013 11:06

1 Attachment(s)
I have been trying to use ANSYS FLUENT Software to simulate cold flow analysis inside a trapped vortex combustion chambustor.My geometry is in 2D-Axisymmetrical. I used pressure based transient case for that. i used k-w sst turbulance model.
And boundry condition is:
Inlet: velocity=40m/s
turbulent intensity :4.5% and inlet duct height:5mm
Outlet:pressure outlet: gauge pre:0

but problem was i got x-momemtum error

plese help me for that.

FMDenaro August 17, 2013 11:56

Quote:

Originally Posted by JAY ADHVARYU (Post 446308)
I have been trying to use ANSYS FLUENT Software to simulate cold flow analysis inside a trapped vortex combustion chambustor.My geometry is in 2D-Axisymmetrical. I used pressure based transient case for that. i used k-w sst turbulance model.
And boundry condition is:
Inlet: velocity=40m/s
turbulent intensity :4.5% and inlet duct height:5mm
Outlet:pressure outlet: gauge pre:0

but problem was i got x-momemtum error

plese help me for that.

Sorry, I do not understand the relation between your question and this post...


All times are GMT -4. The time now is 13:06.