Integral Length Scale - 2D

WhiteShadow · May 15, 2018, 07:05

Hello Everyone,

I want to calculate the integral length scale directly from the energy spectrum. According to Wiki (https://en.wikipedia.org/wiki/Integral_length_scale)
this can be done via

$l=\frac{\int_0^\infty E(k)k^{-1}\,\text{d}k}{\int_0^\infty E(k)\,\text{d}k}.$

But I am working on 2D-flows and I have reason to suspect, that this formula should include a factor of 2 in this case, though I couldn't find anything about this.
Does anyone know whether there is indeed a factor of 2 or if I am mistaken?

Thanks a lot and greetings

FMDenaro · May 16, 2018, 03:09

What is your specific reasoning for the factor 2 in 2D?

Actually, my opinion is that any evalution of the integral lenght is for the order of magnitude, so that a factor of 2 would not change so much.
See https://www.researchgate.net/publica...Numerical_Data

WhiteShadow · May 16, 2018, 03:48

I used a gaussian Spectrum with

$E(k)=\frac 18 \sigma^2 k^3 L^4 \text{exp}(-\frac{k^2L^2}{4})$ .

I calculated the longitudinal correlation function via

$R_{11}(r \vec{e}_1)=\int_{R^2}e^{i \vec{k} \cdot \vec{e}_1 r}\Phi_{11}(\vec{k}) \, \text{d}\vec{k}$ .

In this case I obtained

$R_{11}(r \vec{e}_1)/v_{\text{rms}}^2= f(r) = e^{-\frac{r^2}{L^2}}$ .

The integral length scale

is the integral over the correlation function

$l=\int_0^\infty f(r) \,\text{d}r = \sqrt{\pi}/2 L$ .

But if I calculate the same quantity via the method in my first post,
I get

$l = \sqrt{\pi}/4 L$ .

LuckyTran · May 17, 2018, 22:52

These formulae look quite arbitrary. What is the result for 3D?

WhiteShadow · May 18, 2018, 10:46

I'm sorry, but I do not understand. What do you think is arbitrary?

FMDenaro · May 19, 2018, 16:56

I think you could start by the general definition in terms of the integral of the autocorrelation, considering that spectra and correlation are related each other by direct/inverse Fourier transformation.
But in general the determination of the integral scale is somehow still debated, as you can see for example here
http://www.turbulence-online.com/Pub...apers/WG02.pdf

May 15, 2018, 07:05	Integral Length Scale - 2D	#1
WhiteShadow New Member Join Date: May 2017 Posts: 10 Rep Power: 8	Hello Everyone, I want to calculate the integral length scale directly from the energy spectrum. According to Wiki (https://en.wikipedia.org/wiki/Integral_length_scale) this can be done via $l=\frac{\int_0^\infty E(k)k^{-1}\,\text{d}k}{\int_0^\infty E(k)\,\text{d}k}.$ But I am working on 2D-flows and I have reason to suspect, that this formula should include a factor of 2 in this case, though I couldn't find anything about this. Does anyone know whether there is indeed a factor of 2 or if I am mistaken? Thanks a lot and greetings

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
to determine Integral length scale in Fluent	raunakjung	FLUENT	3	February 7, 2017 09:03
Turbulence length scale and integral length scale	rizhang	CFX	2	April 22, 2016 07:22
Integral length length scale	nasir.iitd	FLUENT	2	June 11, 2014 08:37
Integral Length Scale vs Turbulent Length Scale	Scott Nordsen	FLUENT	2	January 20, 2014 01:17
Von Karman Integral Length Scale	Txingurri	Main CFD Forum	0	May 2, 2002 12:05

May 16, 2018, 03:09		#2
FMDenaro Senior Member Filippo Maria Denaro Join Date: Jul 2010 Posts: 6,599 Rep Power: 70	What is your specific reasoning for the factor 2 in 2D? Actually, my opinion is that any evalution of the integral lenght is for the order of magnitude, so that a factor of 2 would not change so much. See https://www.researchgate.net/publica...Numerical_Data

May 16, 2018, 03:48		#3
WhiteShadow New Member Join Date: May 2017 Posts: 10 Rep Power: 8	I used a gaussian Spectrum with $E(k)=\frac 18 \sigma^2 k^3 L^4 \text{exp}(-\frac{k^2L^2}{4})$ . I calculated the longitudinal correlation function via $R_{11}(r \vec{e}_1)=\int_{R^2}e^{i \vec{k} \cdot \vec{e}_1 r}\Phi_{11}(\vec{k}) \, \text{d}\vec{k}$ . In this case I obtained $R_{11}(r \vec{e}_1)/v_{\text{rms}}^2= f(r) = e^{-\frac{r^2}{L^2}}$ . The integral length scale is the integral over the correlation function $l=\int_0^\infty f(r) \,\text{d}r = \sqrt{\pi}/2 L$ . But if I calculate the same quantity via the method in my first post, I get $l = \sqrt{\pi}/4 L$ .

May 17, 2018, 22:52		#4
LuckyTran Senior Member Lucky Join Date: Apr 2011 Location: Orlando, FL USA Posts: 5,496 Rep Power: 64	These formulae look quite arbitrary. What is the result for 3D?

May 18, 2018, 10:46		#5
WhiteShadow New Member Join Date: May 2017 Posts: 10 Rep Power: 8	I'm sorry, but I do not understand. What do you think is arbitrary?

May 19, 2018, 16:56		#6
FMDenaro Senior Member Filippo Maria Denaro Join Date: Jul 2010 Posts: 6,599 Rep Power: 70	I think you could start by the general definition in terms of the integral of the autocorrelation, considering that spectra and correlation are related each other by direct/inverse Fourier transformation. But in general the determination of the integral scale is somehow still debated, as you can see for example here http://www.turbulence-online.com/Pub...apers/WG02.pdf