averaging in les

klilla · February 5, 2010, 04:21

Dear All,

I made a compressible LES simulation and used the averaging function to get the mean and r.m.s. pressure. A bit disturbing, but the r.m.s. pressure contains negative values in some cells. How can this be?

Thanks,

Lilla

eugene · February 9, 2010, 05:02

I don't think the Prime2Mean values are being calculated in the way you think, i.e

Field_rms = sqrt(Mean(sqr(Field-FieldMean)))

Remember, this is a running average, so the initial value for FieldMean is not accurate. As a result the mean squares are calculated as:

Field_ms_new = alpha * Field_ms_old + beta*sqr(Field) - sqr(FieldMean_old) + sqr(FieldMean_new)

where alpha = (N-1)/N and beta = 1/N, which can lead to negative values for small N.

It could also happen, that you restart the calculation with a FieldMean present, but with no Field_ms. This could lead to large initial negative values, as the initial Field_ms is calculated as:

Field_ms_init = sqr(Field) - sqr(FieldMean)

A better initialisation would probably be:

Field_ms_init = sqr(Field - FieldMean)

klilla · February 10, 2010, 03:29

Dear Eugene,

thanks for your answer. Indeed, I checked the implementation as well, and the problem can arise from my test case. I have a pulsating source around a constant flowfield. The source term I guess makes harder to compute the mean, so I decided to fix the mean flow by myself and than calculate p^2 for noise computation.

Best regards,

Lilla

February 5, 2010, 04:21	averaging in les	#1
klilla Member Kapa Lilla Join Date: Mar 2009 Location: Bruxelles, Belgium Posts: 57 Rep Power: 17	Dear All, I made a compressible LES simulation and used the averaging function to get the mean and r.m.s. pressure. A bit disturbing, but the r.m.s. pressure contains negative values in some cells. How can this be? Thanks, Lilla

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
spanwise averaging for LES backward facing step	Sungho Yoon	CFX	6	July 23, 2008 11:34
Averaging LES	iko	FLUENT	3	July 3, 2008 06:55
Start of averaging in LES	fab	Main CFD Forum	2	June 16, 2005 05:53
Some Questions about LES.	Bin Li	Main CFD Forum	2	February 20, 2004 10:58
Averaging Method for LES	Kim	Main CFD Forum	0	November 21, 2001 08:18

February 9, 2010, 05:02		#2
eugene Senior Member Eugene de Villiers Join Date: Mar 2009 Posts: 725 Rep Power: 21	I don't think the Prime2Mean values are being calculated in the way you think, i.e Field_rms = sqrt(Mean(sqr(Field-FieldMean))) Remember, this is a running average, so the initial value for FieldMean is not accurate. As a result the mean squares are calculated as: Field_ms_new = alpha * Field_ms_old + beta*sqr(Field) - sqr(FieldMean_old) + sqr(FieldMean_new) where alpha = (N-1)/N and beta = 1/N, which can lead to negative values for small N. It could also happen, that you restart the calculation with a FieldMean present, but with no Field_ms. This could lead to large initial negative values, as the initial Field_ms is calculated as: Field_ms_init = sqr(Field) - sqr(FieldMean) A better initialisation would probably be: Field_ms_init = sqr(Field - FieldMean)

February 10, 2010, 03:29		#3
klilla Member Kapa Lilla Join Date: Mar 2009 Location: Bruxelles, Belgium Posts: 57 Rep Power: 17	Dear Eugene, thanks for your answer. Indeed, I checked the implementation as well, and the problem can arise from my test case. I have a pulsating source around a constant flowfield. The source term I guess makes harder to compute the mean, so I decided to fix the mean flow by myself and than calculate p^2 for noise computation. Best regards, Lilla