Hi all,

i have an instantanious velocity profile froma  jet cross section (1D) i have subtracted from this the convectional velocity of the profile (the mean) and the result is a profile of velocity fluctuation in the u direction at a given distance.  (data achieved by PIV technique)

The distance array has a 256 points (0.1072 e-3m to 0.2185m) the velocity profile also there for has the same number of points.  the change in d between samples is 0.854mm

i am using the fft of the velocity data myltiplied by the conjugate fft of the velocity data to only deal with the real parts.

ftu1 = fft(u1)*conj(fft(u1))

my x axis is calculated using,

n=length(dmm);  % dmm is imported data array
L= 218.5e-3;
k= [1/L : 1/L : n/L];

when i 
plot (k, ftu1,'.'); 

 i find a very large peak in the first bin,only after this peak (from the second bin to the end) is the data symmetrical.  i am unsure how and why this peak is there.  i am unsure weather this is due to to any spill overs or roll off (decrement). i am pretty knew to using faourier transforms.   i know matlab uses the entire length(x) as the default window function. 

is it valid to simply remove this first data point or pad it with zero's.  since this first peak is not witnessed on the other side of the plot (other side of symmetry) i cannot determine this to be my dominant wave number.  

can anyone please help with this problem. 

Kind regards.
Neet.

On Nov 2, 9:46 am, "Neet Lad" <nits...@hotmail.com> wrote:
> Hi all,
>
> i have an instantanious velocity profile froma  jet cross section (1D) i =
have subtracted from this the convectional velocity of the profile (the mea=
n) and the result is a profile of velocity fluctuation in the u direction a=
t a given distance.  >(data achieved by PIV technique)

If you subtract the mean, the spectrum should be zero
at zero frequency.

> The distance array has a 256 points (0.1072 e-3m to 0.2185m) the velocity=
 profile also there for has the same number of >points.  the change in d be=
tween samples is 0.854mm

Your numbers are not consistent

N    =3D 256
dmin =3D 0.1072e-3
dmax =3D 0.2185
dd   =3D 0.854e-3

(dmax-dmin)/(N-1) =3D 0.85644e-3  % Not 0.854e-3
dmin+dd*(N-1)     =3D 0.21788     % Not 0.2185
(dmax-dmin)/dd     =3D 255.73     % Not 256

> i am using the fft of the velocity data myltiplied by the conjugate fft o=
f the velocity data to only deal with the real parts.
>
> ftu1 =3D fft(u1)*conj(fft(u1))

No.

 That is the sum of the squares of the real
and imaginary parts. It seems to be a measure
of relative kinetic energy density.

if
U1 =3D fft(u1);
then
ftu1 =3D abs(U1).^2; % Not real(U1)

> my x axis is calculated using,
>
> n=3Dlength(dmm);  % dmm is imported data array
> L=3D 218.5e-3;
> k=3D [1/L : 1/L : n/L];

No.

x =3D dmin+dd*(N-1);
L =3D N*dd          % max(x)+ dd
dk =3D 1/L
k =3D dk*(0:N-1)


> when i
> plot (k, ftu1,'.');
>
>  i find a very large peak in the first bin,only after this peak (from the=
 second bin to the end) is the data symmetrical.  i am unsure how and why t=
his peak is there.

It corresponds to the energy of the mean velocity
which should have been subtracted out.

>i am unsure weather this is due to to any spill overs or roll off (decreme=
nt). i am pretty knew to using faourier transforms.   i know matlab uses th=
e entire length(x) as the default window function.
>
> is it valid to simply remove this first data point or pad it with zero's.=
  since this first peak is not witnessed on the other side of the plot (oth=
er side of symmetry) i cannot determine this to be my dominant wave number.
>
> can anyone please help with this problem.

You are doing some thing wrong

1. Your distance numbers are not consistent
2. Your 0 wavelength spectrum value should be zero
if you subtracted the mean velocity

Hope this helps.

Greg

On Nov 3, 2:09=A0am, Greg Heath <he...@alumni.brown.edu> wrote:
> On Nov 2, 9:46 am, "Neet Lad" <nits...@hotmail.com> wrote:
>
> > Hi all,
>
> > i have an instantanious velocity profile froma =A0jet cross section (1D=
) i have subtracted from this the convectional velocity of the profile (the=
 mean) and the result is a profile of velocity fluctuation in the u directi=
on at a given distance. =A0>(data achieved by PIV technique)
>
> If you subtract the mean, the spectrum should be zero
> at zero frequency.

at zero wavenumber.

Hope this helps.

Greg