CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   Sampling of statistical data (http://www.cfd-online.com/Forums/main/11866-sampling-statistical-data.html)

 Boerge July 25, 2006 04:40

Sampling of statistical data

Hi!

I'm a beginner in CFD, and I need to get up to speed on this topic/art fast. I'm redoing the DNS simulation done by Kim, Moin, Moser (J. Fluid Mech. (1987) vol. 177, pp. 133-166) using Fluent 6.2.16 and the LES turbulence model.

My question is simply the following: What is the proper way to sample statistcal data from a CFD simulation?

A brief summary of the simulation case: It is a simulation of the Poiseuille flow (pressure driven flow between two infinte planes). The reynolds number based on the height H between the plates is ~5000. The computational domain is 6H long and 3H wide, periodic conditions in the spanwise and streamwise directions.

I run the simulation for some residence times (10) such that the flow is at a statistically steady state before the sampling procedure starts. I've asked google and taken a quick look in some turbulence books, neither could provide infomation on how data should be sampled. The quantites of interest are mean velocities and rms values.

I have gotten an advise from an another PhD student for computing the mean velocities which is the following:

1) make a spatial average of the quantites of interest, say the streamwise velocity u(y), where y is the distance from the wall.

2) sample in time by doing 1) for each large eddy turnover time.

Fluent can also sample statistical data for me, but then I don't know how the sampling is done. My guess is that Fluent samples for each time step, and I'm not sure that this is the best way of doing it. And how Fluent computes the rms values is unknown.

I'd be very happy to get some advise on how I should do this. Litterature references are also welcome.