Acoustic Flow Measurements and technology

D. Hurther, LEGI Grenoble: General introduction, Acoustic

flow measuring techniques (Flow discharge, ADV,
ADVP,UVP, ADCP)

Environmental Fluid Mechanics

 Lecture content
1. Overview of Acoustic flow measuring systems and applications

2. Acoustic Doppler Velocimetry (ADCP, ADV, ADVP, UVP technologies)

2.1. The Doppler effect

The monostatic case
The bistatic case
The multi-component case: multi-bistatic
The multi-component case: multi-monostatic
2.2. Pulsed Doppler velocity systems for Local (point-wise)
velocity measurement
2.3. Pulsed Doppler Velocity for velocity profiling
The monostatic case
The bistatic case
2.4. (Pulse-to-pulse) Coherent Doppler velocity systems
The local (point-wise) case
The profiler case
Velocity / profiling-depth ambiguity relation
2.6. Doppler noise & noise reduction methods
 Lecture content

2.7. Acoustic sensor characteristics

2.8. De-aliasing techniques
2.9. Hardware setup and functionning
System components
Doppler signal demodulation
Doppler frequency estimator
Sensor characteristics

3. Acoustic Backscattering for particle concentration & particle size measurements

3.1. Basic theory of sound propagation
3.2. Sound scattering from a sphere
3.3. Sound scattering from a sediment particle
3.4. Sound scattering from a suspension of spheres
3.5. Sound scattering from a sediment suspension
3.6. Inversion techniques for concentration & size calculations
 Lecture content

4. Acoustic systems for flow discharge measurements

4.1. Acoustic flow entrainment effect
4.2. Phase shift based systems
4.3. Frequency shift based systems
4.4. Transit time based systems
4.5. Beam deviation based systems
1. An overview of Acoustic flow
measuring systems and
applications
 An overview of acoustic flow measuring systems
•  Acoustic transit time systems:
Atmospheric flow application:

Average Wind velocity measurement on a

meteorological tower, between an emitter and
receiver: in this case 2C orthogonal velocity
components can be measured at a rate of several
5cm Hz

River flow application:

Average (along a line) flow velocity measurement

between an emitter and receiver: measurement is
done at several flow depths in order to estimate
the water discharge. Continuous flow discharge
suveys by hydro-electric companies,

6
 An overview of acoustic flow measuring systems
•  Acoustic Doppler Current Profiler measurements (based on ac. scattering)

Large river / estuary / ocean

applications:
Long range (up to 500m) 3C current
velocity profiling (1D) along flow
depth or other direction (see below).
The maximal resolution in the
profiling direction is about 25cm.

Continuous flow discharge and sediment

transport surveys in large rivers :
In this case the H(orizontal)-ADCP is profiling in
crossflow direction for 3C1D velocity and
sediment concentration profilings. ADCP
technology has become a widely used
technology for industrial, research, water
resources management all over the world. 7
 An overview of acoustic flow measuring systems
•  Ac. Doppler Velocimeter / Ac. Doppler Velocity Profilers / Ac. Backscattering Syst.
ADV pointwise meas.
Gravel-bed channel
flow

5cm

ADV Profilers

High-resolution velocity and sediment concentration

measurements:
Point-wise (ADV) or profile (ADVP) measurement of 1C to 3C
velocity at small turbulent flow scales. Used to study
fundamental flow and sediment transport processes in lab and
field environment.
Source: P. D. Thorne

Both principles and techniques are recently combined in a single system: Acoustic Concentration
8
and Velocity Profiler (ACVP): very powerful tool for sediment transport meas.
-Acoustic flow measuring system work in the ultrasound frequency
range f>25KHz

-rely on the following physical principles:

•  sound entrainment by the flow
•  Doppler effect
•  Rayleigh to geometric sound scattering
•  viscous sound absorption (by fine suspended matter)
•  sound attenuation
•  multiple sound scattering

-Acoustic flow measuring systems provide flow discharge, up to 3C

velocity meas., suspended sediment concentration, suspended
sediment size, sediment flux, bed and bedform tracking

-robust and versatile for use in both laboratory AND field conditions
in opaque liquids and under extreme flow forcing situations (storms,
floods, dam flushing,…)
9
 2. Acoustic Doppler Velocimetry
(ADV, ADVP, ADCP)

2.1. Doppler effect

Let’s consider a flow entraining fine suspended particles (impurities, micro-air bubbles, fine
 
sediments, biologic particles) moving at speed V = V p ( u,v ) . Due to a variation of acoustic
impedance between the flow medium and the particles, an emitted sound wave
propagating across the flow medium will be scattered by the moving targets. Doppler
Velocity systems generate ultrasound waves and acquire this scattered ultrasound wave to
€  
extract the Doppler frequency and associated velocity. When V ≠ V p ( u,v ) , the measured
velocity corresponds to the particle / sediment velocity which is of great interest for
particle / sediment transport studies.
€
 Doppler frequency in monostatic configuration

Incident pressure wave at E: E/R j
Definitions:
  ⎛ u t ⎞
 
rp = V p t = ⎜ ⎟
⎝v t ⎠ ei i
 2π  ⎛0 ⎞
Pressure wave at target particle P: ki = ei = ⎜ ⎟ €

   λ ⎝ −2π λ ⎠
 V p ( u,v )
[ (
pT = bcos ω 0 t + k i ⋅ rE − rp )]
€

€

ks =
2π  ⎛0
λ
es = ⎜
⎞
⎟
⎝2π λ€
⎠
es
€
P
Received pressure wave at R=E:
    €     € 
€  
[ ( ) (
pR = bcos ω 0 t + k i ⋅ rE − rp + k s ⋅ rp − rR )] [ ( )
= bcos ω 0 t + rp ⋅ k s − k i + k i ⋅ rE − k s ⋅ rR ]
     
[
= bcos 2π ( f 0 + f D )t + φ ] φ = ki ⋅ rE − k s ⋅ rR = const. for VR = VE = 0
The Doppler frequency at the fix point R=E (emitter and receiver sensor position 
monostatic configuration) is the difference between the received frequency and the
incident frequency: €
1 d    f0   
fD = fR − f0 =
2π dt [ (
rp ⋅ k s − k i =
c
V p ⋅ (e s − e i ))] [ ]
f0   2f
fD = 2(V ⋅ es ) = 0 v
c c
€
 Doppler frequency in bistatic configuration
Definitions:
  ⎛ u t ⎞
rp = V p t = ⎜ ⎟ E
Incident pressure wave at E: ⎝v t ⎠
 2π  ⎛0 ⎞  R
ki =
λ
ei = ⎜ ⎟
⎝ −2π λ ⎠
ei
€  2π  ⎛ sin α ⎞ 
Pressure wave at target particle P: ks = es = 2π λ ⎜ ⎟ es
  
λ ⎝ cos α ⎠ 
€  V p ( u,v )
[ (
pT = bcos ω 0 t + k i ⋅ rE − rp )] j
€
€  P
Received pressure wave at R≠E: i
       €     
[ ( )
pR = bcos ω 0 t + k i ⋅ rE − rp + k s ⋅ r€ (
p − rR )] [ ( )
= bcos ω 0 t + rp ⋅ k s − k i + k i ⋅ rE − k s ⋅ rR
€ ]
     
[
= bcos 2π ( f 0 + f D )t + φ ] = ki ⋅ rE − k s ⋅ rR = const.
φ€ for VR = VE = 0

The Doppler frequency at the fix point R≠E (emitter and receiver sensor are different
 bistatic configuration) is the difference between the received frequency and the
€
incident frequency:
1 d    f0   
fD = fR − f0 =
2π dt [ (
rp ⋅ k s − k i )] =
c
[
V p ⋅ (e s − e i ) ]
f0
fD =
c
[ usin α + v (1+ cos α )]
€
 Multi-component (multi-bistatic) Doppler velocity measurement
When several velocity components have to be measured simultanously, several
acoustic sensors are placed in the flow to form a multi-bistatic or multi-monostatic
configuration
 
•  Let’s consider a 2D flow with V = V p ( u,v )
that has to be measured using a
symmetrical two-bistatic system composed of one central Emitter at point E and two
receivers at points R+ and R-, as seen below
€
  ⎛ u t ⎞  2π  ⎛0 ⎞
Definitions: rp = V p t = ⎜ ⎟ ki = e = ⎜ ⎟
⎝v t ⎠ λ i ⎝ −2π λ ⎠

€ €
 2π  ⎛ sin α ⎞
ks + = e + = 2π λ ⎜ ⎟
λ s ⎝ cos α ⎠
 2π  ⎛ −sin α ⎞
ks − = e − = 2π λ ⎜ ⎟
λ s ⎝ cos α ⎠
€
v
13
€
 Multi-component (multi-bistatic) Doppler velocity measurement

  ⎛ u t ⎞  2π  ⎛0 ⎞
rp = V p t = ⎜ ⎟ ki = e = ⎜ ⎟
⎝v t ⎠ λ i ⎝ −2π λ ⎠

 2π  ⎛ sin α ⎞
ks + = es + = 2π λ ⎜ ⎟
€ λ € ⎝ cos α ⎠
 2π  ⎛ −sin α ⎞
ks − = es − = 2π λ ⎜ ⎟
λ ⎝ cos α ⎠ v
€

1 d    f
€ f D+ =
2π
r ⋅
dt p
( )
ks + − k i = 0 [ usin α + v (1+ cos α )]
c
Doppler frequencies
1 d    f
f D− =
2π
r ⋅
dt p
( )
ks − − k i = 0 [ −usin α + v (1+ cos α )]
c
€ ⎧ c
⎪
⎪
u =
2 f 0 sin α
(
f D+ − f D− ) Most of existing ADV and ADVP systems
Multi-Bistatic 2C velocity ⎨ use this technique. The two Doppler
€
reconstruction ⎪v = c frequencies are measured at the same
⎪
⎩ 2 f 0 (1+ cos α )
(
f D+ + f D− ) point along the common emitter axis
Add 3c case of Multi-bistatique ADV
 Multi-component (multi-monostatic) Doppler velocity
Definitions:
E- / R- E+ / R+ 
  ⎛ u t ⎞
− + j
rp = V p t = ⎜ ⎟
⎝v t ⎠ ei ei 
 2π  − ⎛sin α ⎞ − + i
€
k −i =
λ
e i = 2π λ ⎜
⎝ −cos α ⎠
⎟ es es 
€

k −S =
2π  − ⎛ −sin α ⎞
e s = 2π λ ⎜ ⎟ V p ( u,v )
λ ⎝ cos α ⎠ €

k +i =
2π  + ⎛ −sin α ⎞
e i = 2π λ ⎜ ⎟ € €P
λ ⎝ −cos α ⎠
€ 
k +S =
2π  +
λ
⎛ sin α ⎞
e i = 2π λ ⎜ ⎟
⎝ cos α ⎠
€ € €

€
ADCP systems use this technique BUT
Multi-monostatic 2C The two Doppler frequencies are
velocity reconstruction measured at separate points along the
two different emitter axis (except the
intersection point).

This requires flow uniformity assumptions along the two axis of measurement
2.2. Pulsed Doppler velocity systems
for local (point-wise) Doppler velocity
measurement
(ADV)

A Point-wise velocity measurement means that the estimated velocity can be considered
as local with a negligibly low spatial averaging effect of the velocity over the sample
volume. This implies that the size of the sample volume over which the velocity is spatially
averaged is smaller than the smallest flow velocity scale to be resolved. To achieve this,
the emitted ultrasound wave must be time limited as a pulse or pulse train with a duration
beeing smaller or equal to the smallest temporal flow scale to resolve.

17
Justification of acoustic frequency, pulsed
transmission mode & pulse time-length
Pulsed Doppler velocity systems for local (point-wise)
velocity measurement
Let’s consider a monostatic ADV sensor placed at the
 wall of a pipe water flow (diameter D) and at an angle
Emitter /  es
Receiver ei β from the horizontal streamwise flow direction.
: acoustic frequency
€ : sound speed in water (density
€ Kg.m-3, compressibility in Pa-1)
: acoustic wavelength

A short pulse train at frequency f0 is

transmitted into the water to propagate
along the sensor axis at an angle β.
Immediately after emission, the sensor
is switched into receiving mode to
capture the acoustic wave
backscattered by the suspended
acoustic targets advected by the flow.
This cycle is repeated at the Pulse
Repetition Frequency (PRF) until
sufficient samples are collected to form
the Doppler signal
 Pulsed Doppler velocity systems for local (point-wise)
velocity measurement
The following relation are deduced:

: along axis pulse width

: along axis sample volume

width

: along axis distance between

emitter and sample volume
position

: vertical distance between

emitter and sample volume
 tr varies over the range [te ; (TPRF-
: along axis maximal measurement depth tm)] to cover the vertical domain [lesinβ ;
(D-dmsinβ)]
To measure at any point across the  maximal resolution is achieved
pipe, the TPRF must be set such that
with: tm=te

 2.3. Doppler Velocity profiling
(ADVP, ADCP)
Profiling means that the local (point-wise) velocity measurement is performed at several
points distributed along an axis defined by the sensor insonification direction.

Consequently, profilers provide 1C1D to 3C1D velocity measurements dependent on the

number of components to be measured (see section 2.2). This performance is of great
significance for many scientific and engineering problems.

21
 Doppler Velocity profiling (monostatic configuration)
M measurement points in one
profile located along the sensor
axis at a distance dj from the
sensor:

In vertical distance:

Doppler frequency z2: Doppler frequency z1:

  
v r = V ⋅ (es − ei )

Because c=1480 m/s in water >>> V(u,v), typically 103 times larger, the profiling is done quasi-22
instantanously at the scale of the flow €
 Doppler Velocity profiling (multi-bistatic configuration)
Let’s consider a multi-bistatic system composed of:
•  A central emitter E (with high directivity) placed vertically to the streamwise direction
•  An upward oriented receiver R- (with a low directivity)
•  An downward oriented receiver R+ (with a low directivity)

Geometry at gate j in the profile

•  The geometrical parameters D0 and angle γ are known

•  As for the monostatic profiling, the backscattered signal is sampled between two
consecutive emissions in order to measure a velocity profile
•  Because the sensor positions rE ≠ rR+ ≠ rR-, the relation between dj=f(tj) is different 23
from
the linear one valid for a monostatic system.

 Doppler Velocity profiling (multi-bistatic configuration)
Geometry at gate j in the profile

with
v

Knowing the geometrical variables D0, γ and the pulse travel distance ctj=dj+bj for each j.,
the distance dj between the emitter and the measurement point j in the profile becomes:

The local Doppler angle at point j (necessary for the Doppler frequencies calculations) in
the profile is calculated as

24
 Doppler Velocity profiling (multi-bistatic configuration)

•  Linear relationship between dj and tj for

monostatic systems

•  Non-linear relationship for a bistatic

system converging asymptotically to a
linear relation for large sound wave
travelling distances

The spatial resolution defined as:

Is much lower for bistatic systems in the

domain close to the emitter

25
Monostatic and Multi-bistatic Profiler Multi-bistatic Pointwise
Multi-monostatic
Profiler
ADVP systems ADV systems
UVP / ADCP
systems

26
 2.4. (Pulse-to-pulse) Coherent
Doppler velocity systems
(ADV, ADVP)

Because point-wise Doppler velocity systems (ADV) and Doppler velocity profilers (ADVP)
require the emitting of short period pulse trains, the emitting / receiving cycle has to be
repeated coherently in time to recover a long sample of the Doppler signal.
These pulses are usually much shorter in time than the period of the Doppler frequency, as
a consequence one echoe (one reception at point j) can be seen as an instantaneous point
in time of the Doppler wave signal.

27
The local (pointwise) case

From Hurther (1995)

Consequently:
-PRF: the Pulse repetition Frequency is
the sampling frequency of the Doppler
signal
-The Doppler signal is approximated as a
time discretized signal:
 The profiler case
Consequently:
-PRF: the Pulse repetition Frequency is
the sampling frequency of the Doppler
signal
-The Doppler signal is approximated as a
time discretized signal:

29
The Doppler signal is reconstructed coherently from pulse to pulse
 Velocity / profiling-depth ambiguity relation
f0    f0
1. The general form of the Doppler frequency is f D = V ⋅ (e s − e i ) = vr
c c

•  Pulsed Doppler Velocity systems induces that the

Doppler signal is time discretized at the PRF rate
€ with

•  Applying the Shannon’s sampling theorem, induces:

Consequently Aliasing appears at

2. Moreover, for pulsed monostatic systems, the Profiling depth is

3. Introducing the last relation gives the Maximal velocity / Depth ambiguity relation as

The ambiguity resides in the fact that the maximal

absolute velocity (at which aliasing appears) is
monostatic inversely proportional to the profiling depth
because the PRF is the sampling frequency of the
Doppler signal
 Velocity / Depth ambiguity relation

•  For a bistatic system, we have:

•  The Depth / velocity amibiguity relation for a

bistatic pulse-coherent Doppler system becomes

•  When aliasing can not be avoided by system parameter setting (PRF, bistatic geometry,
acoustic frequency reduction), de-aliasing techniques have to be applied to the Doppler
signal.
2.6. Doppler noise and noise reduction
methods
 Doppler noise and noise reduction methods A.33

•  Noise affected velocity signal:

: spatially averaged over sample vol. : Geom. Reconstruction

with T, transformation
: Turb. Velocity variance matrix

•  Noise affected velocity variances:

with

•  Redundant velocity measurement :

33
 Doppler noise and noise reduction methods
•  Hurther and Lemmin (2001), noise reconstruction

Valid for a symmetrical dual-bistatic system

•  Garbini (1982) proposed

•  Turb. Velocity variance generated by isotropic scales smaller than Δz is also eliminated
•  Hurther and Lemmin (2007): Bi-frequency ADVP allowing velocity redundancy at two carrier
frequencies with a single bistatic.
Turb. Spectra (Hurther and Lemmin 2007) TKE profile (Hurther and Lemmin 2007)

34
2.7. Acoustic sensor characteristics
 Flat piston working piezo-electrical transudcer
Transducers ultrasonic field

In Doppler velocimetry, it usually aimed to generate an ultrasonic beam that is as thin as possible throughout the
measurement profile in so that small flow velocity scales can be resolved. The geometry of the acoustic field is
governed by the radius at of the emitter and the wavelength of the ultrasonic waves λ.

  Pressure radiation in polar coordinates by a flat

disc shaped transducer of radius at

Sensor directivity

First order Bessel function

 Nearfield Z along sensor axis can be

approximated with

In the near field, the acoustic field is basically cylindrical, with a diameter slightly less than the diameter of the
emitter, and the intensity of the acoustic waves oscillates along the axis of the transducer. The length of the near
field, Z, corresponds to the position of the last maximum of the acoustic intensity. Due to the important intensity
variations in this field, Doppler velocity measurement can be affected and are usually avoided there.
 Flat piston working piezo-electrical transudcer
 In the farfield, the pressure amplitude along
sensor axis due to spherical spreading is

In the far field, the pressure amplitude of the acoustic waves along the axis varies as the inverse of the distance from the
transducer and small oscillations appear in the radial direction. Most of the acoustic energy is contained in the main lobe
(conical shape) of which the half angle d is characterized by the wavelength and the radius of the emitter.

  Main lobe beam spreading at first zero in transverse direction at a given axial distance:

  at -3dB attenuation
 Beam spreading diagram (with sensor Diameter) for plain disc transducer in water at 25°C (c=1500m/s)
 Beam spreading decreases with frequency for a given Diameter

 Beam spreading decreases with Diameter at a given frequency

 BUT: Nearfield increases with frequency and Diameter: COMPROMISE to

be found

 Sample volume for pulsed systems (ADV, ADVP, ADCP, UVP,…) can be
approximated by a cylindrical shaped volume as:
2.8. De-aliasing techniques
 De-aliasing techniques
Let’s consider an ADVP system (pulse-to-
pulse coherent Doppler velocity profiler)
deployed below coastal waves to measure
the wave velocities. The Doppler frequency
for each receiver is:
f0    f0
fD = V ⋅ (e s − e i ) = vr
c c
The presence of Doppler frequency
jumps of amplitude at sequences A, B, C,
D, E equal to ± PRF is typical of aliasing
€ effects since such events can not be
affected to physical wave velocity
accelerations.
The time serie of fD at R1 shows the oscillatory cycles of the wave velocity:

A B C

D E
 De-aliasing techniques
In the phase plan, an aliased event such as
at point A in the time serie, can be
represented as follows.

Let’s define:

we can write:

The difference of the phases is then

The aliased Doppler frequency at A can
resulting into be corrected by adding PRF. The same
correction can be done at all other jumps
(B, C, D, E)
A
 De-aliasing techniques
The following detection parameters Time evolution of detection parameter
can be calculated as

The corrected Doppler frequency is obtained

using the following optimization criteria

-Can be applied systematically when:

–  Initial fD,2(t=0) is not aliased
–  fD,2 time fluctuation due to Doppler noise
or physics <±PRF/2
-Technologically very easy to implement
compared to dual-PRF methods
 De-aliasing techniques
Based on the fact that the aliased Doppler can be corrected, the following de-aliasing technique can be
proposed:

At each time step two additional

Doppler frequencies and can be
calculated as
2.9. Hardware setup and
functionning
Typical system setup Fluid Hardware
Hardwareunit
unit Software
Softwareunit
unit
Fluid medium
medium
Settings
Active
meas. Control
signal Meas. Data
signal
Signal pre- value A/D tranfser Data storage
Sensor
treatment Digital processing Software post-
Sensor processing
Meas. quantity
output
signal
contineous signals discretized signals

Softw.
unit Hardware unit
Fluid medium
Setup ADV / ADVP systems

Sensors

Hardware unit

Hardware unit Software unit

44
 Doppler signal demodulation

•  Emitter signal at carrier pulsation ω0

•  Received signal (idealized):

•  Complex Doppler signal

with
 Doppler signal demodulation
•  Complex Doppler signal

with

  The complex signal D(t)=Re(t)+i Im(t) is necessary for the determination of the velocity sign
since

 hLP: low-pass filter with f-3dB as the -3 dB cutoff frequency. To filter out the high-frequency
component resulting from the analogue multiplication, the :

  Furthermore, in order to avoid « point-to-point » filtering effects:

Note that for most application tm=te in order to reach the maximal spatial resolution.
46
 Doppler frequency estimator: the pulse-pair algorithm

•  autocorrelation function of complex D(t):

with

  time resolution becomes (Npp x TPRF ), Npp is the number of consecutive echoes
taken to calculate one Doppler frequency fD.

 the estimated Doppler frequency is unaffected by the noise signal n(t) in D(t) if it is
white (never true, needs further noise reduction techniques)

 typically the Doppler noise variance converges for Npp between 8 and 32
depending on the « quality » of the Doppler signal

 Homogeneity of tracer distribution within the sample volume essentially determines

47
the signal quality
48
 General ADVP parameters and resolutions
•  ADVP parameters and range
PRF Pulse Rep. Freq. 300Hz...3KHz
f0 carrier freq. 500KHz…16MHz
te Pulse duration 1µs…20 µs
Ng Points / profile 1…256
Δte Gate time spacing 1µs…20 µs
lm Gate spacing 1.5mm to 50cm
D Profiling depth 10cm to 2m
Nr Number of receivers up to 7
ΔVi Velocity resolution O(1 mm/s)

•  Resolutions (typ. for channel flows applications)

lm

49
 3. Acoustic Backscattering for particle
concentration & particle size measurements