Geoscience Reference
In-Depth Information
where
c
is the speed of sound in the experimental con-
ditions. For a given incident frequency
ν
0
and a given
scatter angle
θ
s
, the instantaneous frequency shift
δν(t)
=
ν
s
−
open flows with large mean velocities. The last point is
particularly important, for instance, in wind tunnel or
jet experiments where efficient optical tracking generally
requires to mount the camera on a platform moving at the
mean wind speed [
Ayyalasomayajula et al.
, 2006] in order
to track particles for sufficiently long times (typically com-
parable to the largest time scales of the flow) while acous-
tic tracking can be efficiently done using fixed transducers;
moreover, a backscattering configuration (
θ
scatt
ν
0
gives a direct measurement of the projection,
v
//
, of the particle velocity along
n
s
(note that this
is an algebraic measurement: the sign of
v
//
is given by
the sign of the frequency shift). Hence, the continuous
recording of the frequency shift
δν(t)
gives a Lagrangian
measurement of the velocity component
v
//
(t)
along the
particles trajectory. It is interesting to note the analogy
of this acoustic technique with the optical ELDV method
previously described. The modulation of scattered light
by a particle moving in the interference pattern in ELDV
is indeed conceptually equivalent to the modulation of
the Doppler-shifted acoustic wave scattered by a parti-
cle in the present configuration. Finally, we point out
that the combination of several pairs of transducers and
working frequencies allows to extend the measurement
and access two or three components of the velocity (see
Figure 15.8, bottom).
Such an acoustic Lagrangian tracking technique was
first implemented by
Mordant et al.
[2001, 2002, 2005] in
a pioneering study of Lagrangian turbulent statistics in a
von Kármán swirling flow of water. In that case, piezo-
electric elements were used as acoustic transducers (with
typical emitting frequency operating in the megahertz
range) and small polystyrene particles served as tracers.
More recently, the same technique was ported to inves-
tigate opened air flows in a turbulent jet [
Poulain et al.
,
2004] and wind-tunnel experiments [
Qureshi et al.
, 2007,
2008]. It is interesting to note here that the same identical
systems used by Poulain et al. and Qureshi et al. can also
be easily ported to perform in situ measurements in real
atmospheric flows. In these experiments Sell-type acoustic
transducers were used, operated with typical frequencies
around 100 kHz (ultrasounds at higher frequency are
rapidly damped in air) and tracked particles were small
millimeter soap bubbles either neutrally buoyant (bub-
bles are then inflated with helium) to have tracer behavior
or intentionally heavier than air in order to address the
question of the turbulent transport of inertial particles.
Pros and cons of scattering techniques compared to
direct optical methods can be discussed. The main advan-
tage of acoustics (shared with ELDV) is that Doppler shift
measurements give a direct access to the tracer's velocity,
while direct optical tracking requires to differentiate the
position signal of the tracked particle to get velocity, an
operation which is very sensitive to noise. Hence optical
tracking usually requires severe oversampling if velocity
and acceleration statistics are to be investigated. Other
advantages of acoustic tracking are (i) the low cost of the
required equipment, compared to expensive high-speed
cameras; (ii) the possibility to probe flows in opaque fluids
(as liquid metals for instance); (iii) the possibility to easily
explore large volumes; and (iv) the ability to investigate
n
0
−
180
◦
)
allows to significantly extend the streamwise dimension of
the measurement volume [
Qureshi et al.
, 2007, 2008].
Among disadvantages, a strong limitation of acoustic
tracking is its inability to accurately track several particles
simultaneously. If more than one particle is present in the
measurement volume, the signal recorded by the receiver
superimposes the waves scattered by all the particles.
Although signal processing strategies (discussed below)
do exist to extract the contributions from each individual
scatterer, the accuracy decreases with increasing number
of particles.
15.3.1.2. Signal Processing and Doppler Shift Extrac-
tion.
The acoustic signal recorded by the receiver com-
bines a spectral component around the emitting frequency
ν
0
, which corresponds to echoes and reflections directly
incoming in the receiver without being scattered by the
particles and a Doppler-shifted component around
ν
d
resulting from the fraction of acoustic wave scattered by
the moving particle. The information about particle veloc-
ities is entirely encoded in the Doppler shift
δν
=
ν
s
−
ν
0
.
Therefore, a heterodyne downmixing is generally oper-
ated between the emitted and received signal which essen-
tially results in shifting the emitting frequency
ν
0
to zero.
Figure 15.9 shows a typical spectrum of a downmixed
10
-10
10
-12
10
-14
f
c
F
c
10,000
Frequency (Hz)
0
2000
4000
6000
8000
12,00014,00016,000
Figure 15.9.
Typical spectrum of the donwmixed acoustic sig-
nal recorded by the receiver. The Doppler shift resulting from
the scattering by moving tracers is visible around 7 kHz.
From
Qureshi
[2009].
