Velocity profile measurement by ultrasound

Doppler shift method

Y. Takeda

The application of an external blood flowmeter, an ultrasonic Doppler shift

detection device, to the one-dimensional velocity profile measurement of the
general flow of water was studied. Experiments were carried out for Poiseuille flow
in a 1 2 mm diameter pipe and for Taylor vortex flow in a roating double cylinder
with a 5 m m gap. Measured velocity profiles showed good agreement with
theoretical predictions, implying that the device works quite successfully for this
purpose.

Keywords: flow velocity profiles, flowmeters, ultrasonic Doppler shift

Introduction Experimental apparatus

For heat transfer studies, knowledge of flow behaviour is Principle of measurement
obviously very important. Various methods and
techniques have been applied in investigations especially The EBF is a device to measure the time change of the flow
related to nuclear, solar and fire research, where rate of blood by the ultrasonic Doppler shift detection
knowledge of flow patterns in enclosures is required. For method. The principle of the method is illustrated in Fig 1
these purposes flow visualization techniques have mainly for the example of one-dimensional flow in a pipe or
been used, in order to obtain spatial information on flows. parallel plates. Fig 2 is a schematic diagram of its
However, this method has the disadvantage of difficulties electronic circuit. A transducer is placed on a wall with an
in obtaining quantitative results and real time data angle inclination 0 to ensure a velocity component in the
handling. Furthermore, its application to opaque fluids is direction of the measuring line. A transmitter emits an
not possible. ultrasonic pulse, the basic frequency of which is 4.2 MHz
Laser Doppler anemometry is now quite popular and pulse width 2 #s, repetition period 122/~s. Fig l(b)
for fluid flow research. Although this method gives a shows an example of the signal of the received echo.
response time short enough for real time data acquisition, Between two emissions of ultrasonic pulses the echo can
and resolution high enough for time-series information, it be observed, backscattered from somewhere on the
gives only pointwise information, which is of little use for measuring line. A position axis along the measuring line is
studies of spatial structure of flows. Recently, an external thus converted to a time axis. The frequency of an echo
blood flowmeter (EBF) has been developed in the medical signal at any instant is Doppler-shifted due to the fluid
field. This device uses the principle of Doppler shift of motion as far as it is reflected inside the flow. After the
ultrasound to measure the profile of blood flow in a blood amplification of this echo signal, it is demodulated to
vessel, in order to give the time-dependent flow rate of eliminate the component of basic frequency and then
blood. It employs the pulsed echo method, detecting the digitized. The sampling period of this digitization is fixed
Doppler shift in the echo as a function of time after the at 122-128 #s, implying that the maximum measurable
pulse emission. Reflection of the ultrasound comes from length is divided into 128 channels. The time series of the
the interface of blood cells. It gives spatial information on digitized data is then filtered digitally to eliminate
the velocity profile in a blood vessel within a short time, spurious frequency components which may arise from the
and hence the profiles as a function of time. In this work, vibration of the container or from noise, then the Doppler
an attempt was made to apply this device to flow shifts are detected by counting up the zero-crossings for
measurement of general fluid. Two kinds of configuration each channel, to give a histogram of velocity values, and
were used with water. One is a Poiseuille flow in a simple thus the velocity profile (Fig l(c)).
pipe and another is a Taylor vortex flow. The former was In the histogram, the position is determined from
selected because of its simplicity of flow and its well an echographic equation:
established theory. The latter was chosen because of its d=ct/2
somewhat complicated but interesting velocity profile
along the longitudinal direction and because of its where d is the distance from the transducer, c is the speed
experimental simplicity. of ultrasound in the fluid, and t is the reception time.
From the Doppler shift frequency of each channel,
velocity values can be calculated from the following
*Schweizerisches Institut f'fir Nuklearforschung, 5234 Villigen, Doppler equation:
Switzerland V ML c
Manuscript received 20 August 1985 and accepted for publication 16 v,
October 1985 cos 0 - c2fo
os~f, D

Int. J. Heat Et Fluid Flow 0142-727X/86/040313-06$3.00@ 1986 Butterworth 8 Co (Publishers) Ltd 313
Y. Takeda

eliminate the multiple reflection of ultrasound from the

surroundings. The transducer was set at the angle of 60 °
with respect to the flow. Flow rate was measured by
0~nsducer weighing the water quantity.

Wall
amll~
Flow
Transducer
Transmitter ,~I ]

" X Measuringline

r I
Optimized
amplifier
1
B ........ "
g ....... v

Demodulation

ADC
Separation
of echo

Fig 1 Principle of the method

Zero
detector
where V~ is the velocity value at channel i; V/MLis the
velocity component at each channel i in the direction of
the measuring line; c is the speed of ultrasound in the
fluid; f0 is the basic ultrasound frequency; fD is the
DAC
display
]
Doppler frequency shift for channel i; 0 is the angle
between transducer and flow direction.
When the present device is used with water, the
maximum measurable distance from the transducer is
approximately 9.2cm (c=1500m/s), with a position Fig 2 Schematic diagram of electronic circuit
resolution of about 0.7 mm. With inclination angle of 60 °
the maximum measurable flow velocity is about 73.2 cm/s
with a resolution of 0.6cm/s. Since it measures only
Doppler shift, reflections from the interface between fluid
and the container wall do not affect the measured results.
The data acquisition time to give one frame of profile is
approximately 16ms. The size of the transducer used in

1
this work is 4 m m in diameter. Since a prototype
flowmeter was used, which was equipped only with a
DAC output, the profiles obtained were displayed on the
oscilloscope and photographs were taken of its display. 1~12mm
Poiseuille f l o w .am ; I . ~ - -
A schematic layout of the experimental setup is illustrated . _ = _'P_~_ _

in Fig 3. The inner diameter of the pipe is 12 mm and the

entry length to the measuring point is 1.8m. As a
reflecting material, A1203 powder whose particle
diameter size was approximately 5 #m was suspended in v~
water. The part of the pipe at which the profile was
measured was immersed in a water bath in order to have Fig 3 Schematic layout of experimental setup for
good contact between the transducer and the pipe and to Poiseuille flow

314 Vol 7, No 4, December 1986

 Velocity profile measurement by ultrasound Doppler shift method

Taylor vortex f l o w
The experiments were carried out in a rotating double
cylinder. The diameters of the inner and outer cylinders
were 10 and 11 cm, respectively, and the height of the
water layer was 6.0 cm. The outer cylinder was fixed and
only the inner one was rotated. The rotation speed, set by
observing the flow, was such that the Ta number was
definitely beyond the critical value for the production of a
roll structure. The axial velocity profile was measured by
setting the transducer vertically upward on the bottom
face of the container. Two measurements were performed
for one rotation speed, one along the outer surface and
another along the inner.

Experimental results and discussion

Poiseuille f l o w
Fig 5 Same as Fig 4, but smeared for 50 pictures
Fig 4 shows a typical example (Re = 1945) of a measured
profile. The lower line is an echo signal and the upper one
is a velocity profile converted from the Doppler shift
frequency of the echo. The profile has a histogram Re = 2125 Re = 9 0 2
structure due to the digitization of the echo signal. It 40 40
shows a parabolic distribution of velocity. The width is
14.1 mm, which agrees quite well with the path length of
ultrasound in the pipe (13.9mm). However, it was A 30 30
observed that this profile changes its height and its shape
from picture to picture although every picture showed an
essentially parabolic distribution with little change of •g 2o 20
0
width. This amplitude variation and deformation of
profile arise partially because the amount of reflecting
powder is not optimum and partially because of its
density fluctuation. Accumulating 25 to 50 frames of these
pictures leads to a smoothed average of the profile (Fig 5).
The remaining 20°//0fluctuation of the peak velocity in this
case is presumably due to variation in the flow velocity Position, arbitrary units Position, arbitrary units
itself. In this simple experimental arrangement a stirrer
was used to mix the powder uniformly in the top reservoir, Fig 6 Measured profile for Poiseuille flow and parabolic
and no special care was taken to stabilize the water level. fitting
Reading the average voltage values on a picture gives a
numerical velocity profile. Fig 6 shows the profiles for
Re=2125 and 902. These profiles were fitted to a
parabolic function by a least-square method, which from the fitted data are 1539 and 547 with differences of
reproduced the data quite well. The Re numbers derived 28% and 40°/o respectively from the value obtained from
flow rate measurements. Considering the condition of
nonoptimum amount of reflecting particles, and the
diameter of the transducer and pipe, the present method
tends to give lower values of velocity, so that the outer
envelope containing all the fluctuations corresponds
rather to the profile on the plane which goes through the
centre line of the pipe. This leads to Re numbers of 1747
and 728, reducing the difference to 17 and 19%. This is
then in reasonable agreement, in view of the limited
accuracy of the mass flow measurement. The flowrate
measured in this work was an average over a few minutes
whereas the velocity value derived from the profile gives
an instantaneous value. It can be expected that the
fluctuation of flow during those few minutes would cause
the discrepancy in the determination of Re numbers. Fig 7
shows variations of the velocity profile for various Re
numbers. As clearly seen, for Re numbers lower than 2145,
profiles are parabolic but start to deform from Re ---4120.
Especially at Re = 6565 a very strong fluctuation appears
on the tail side of the profile. This may be due to the
Fig 4 Oscilloscope output: echo and Doppler shift profile internal motion of At20 3 powder in the turbulent motion
for Poiseuille flow (Re = 2125) of the fluid and the resulting multiple reflection of

Int. J. Heat f~ Fluid Flow 315

Y. Takeda

ultrasound in the tube. The front part of the profile shows

much less fluctuation. Voltage values of this part of the
profile was read and converted to a velocity profile, and it
was fitted to a power law distribution. The calculated
value of the power was 0.137 and shows quite good
agreement with the 1/7 power law of turbulent velocity
distribution in a pipe for this range of Re number.

Taylor v o r t e x f l o w
Fig 8 is an illustration of a typical vertical cross-section of
the roll structure in a Taylor vortex flow in a rotating

1945
+ ~11 cm

~10 cm

÷
3090

"4"

4120
- "4"

6566
////
f'JJ, B'JJl'JJJJl'JJJ,
"

Trx Trx
(In) (Out)

10006 ......
Fig 7 Comparison of velocity profiles of Poiseuille flow I I
for various Re numbers Fig 8 Roll structure of Taylor vortex

316 Vol 7, No 4, December 1986

 Velocity profile measurement by ultrasound Doppler shift method

double cylinder. Fig 9 is a measured profile of the read and the velocity profiles were reconstructed and
longitudinal velocity for Ta = 2066, where (a) is the outside plotted in Fig 10. This structure is in good agreement wit h
surface of the gap and (b) is the inside. On the echo signal the theoretical calculation of Ref 2. The distance between
the reflection from the top surface is recorded and it can be two crossings of the zero line correspond to the size of o n e
used for calibrating the distance on the picture. roll. It gives an aspect ratio for the cross-section of a roll of
Oscillatory behaviour is clearly obtained. As seen in Fig 8, about 1.2, which also agrees fairly well with Ref 2.
the velocity along one boundary reverses its direction
from roll to roll. Moreover, on both boundaries at the
same height, the velocity has the same amplitude but with Conclusion
opposite signs. The phases of both measured profiles are By making use of the external blood flowmeter, which is a
opposite to each other, which implies that the velocity is stand-alone machine for medical use, the feasibility of the
opposite and demonstrates the roll structure of the flow. application of a Doppler shift ultrasonic technique to
For this Ta number, the rolls are not fully established in velocity profile measurement. It is concluded that, in
the middle part of the system. In this region the velocity principle, the device is applicable for this purpose.
has actually decreased below detection threshold. The advantage of this method is that spatial
However, on the upper and lower part of the system the information on flow behaviour can be obtained
roll structure is well developed. These profiles were also instantaneously. The accuracy of the velocity

Fig 9 Profile of longitudinal velocity for Taylor vortex flow (Ta = 2066): (a) outside of gap; (b) inside

Longitudinal velocity profile : Taylor vortex flow

10

o
t~

rr

-10
Distancefrom bottom
Fig 10 Measured vertical velocity profile of Taylor vortex (Ta = 2066)

Int. J. Heat ~t Fluid Flow 317

Y. Takeda

measurement and the resolution in position and time are Acknowledgements

deeply related to the selection of frequency of ultrasound,
The author expresses his gratitude to Prof Dr Mooser and
its pulse structure and electronic constants for data
Dr J.-J. Meister of Ecole Polytechnique Federale de
acquisition. They can be adapted to the requirements of
Lausanne for offering him the opportunity of performing
various different cases. The limitations of the method arise
this investigation. He also appreciates the help of Mr
first from the fundamental properties of ultrasound,
C. Gerber for preparing and carrying out the experiments
namely that the measurable depth is limited by the
as well as Dr W. E. Fischer and Dr C. Tschalaer for their
attenuation of ultrasound. Furthermore, the method
encouragement.
requires some amount of reflecting particles suspended in
the liquid, which may disturb its flow or change its
fundamental properties. References
Although the device was developed for
1. Meister J. J. Mesure par echographie Doppler et modelisation
measurement of one-dimensional flow, profiles can also theorique de l'effet de troubles cardiaques sur la pression et le debit
be successfully obtained for flow which is essentially arteriels. Th~se No. 504, 1983, Ecole Polytechnique Federale de
multi-dimensional. Considering its principle, it would be Lausanne
promising to apply the same method to opaque fluids 2. Aiziary de Roquefort T. and Grilland G. Computers and Fluids, 1978,
such as liquid metals. 6, 259

1987
Valve technology 26-28 January 1987 The Center for Professional Advancement, Palestrinastraat 1, 1071
Amsterdam, The Netherlands LC Amsterdam, The Netherlands
Pressure vessel design and 16-19 February 1987 The Center for Professional Advancement, Palestrinastraat 1, 1071
analysis Amsterdam, The Netherlands LC Amsterdam, The Netherlands
10th B P M A Technical 24-26 March 1987 BHRA Fluid Engineering Centre, Cranfield, Bedford MK43 0AJ,
conference on pumps Cambridge, UK UK
Large scale applications of heat 25-27 March 1987 BHRA Fluid Engineering Centre, Cranfield, Bedford MK43 0AJ,
pumps Oxford, UK UK
11th international conference 8-10 April 1987 BHRA Fluid Engineering Centre, Cranfield, Bedford MK43 0AJ,
on fluid sealing Cannes, France UK
International meeting on heat 13-17 April 1987 Lyon, Professor J. F. Sacadura, INSA de Lyon, LMFT Bat. 302-20 Ave.
transfer France Albert Einstein, 69621 Villeurbanne Cedex, France
1987 lEA Heat pump 28-30 April 1987 Ms Pamela J. Lewis, Oak Ridge National Laboratory Building
conference Orlando, Florida, USA 9102-2, PO Box Y, Oak Ridge, Tennessee 37831, USA
International conference on 12-14 May 1987 Bowness-on- BHRA Fluid Engineering Centre, Cranfield, Bedford MK43 0AJ,
f l o w induced vibrations Windermere, UK UK
3rd international conference on 18-20 May 1987 BHRA Fluid Engineering Centre, Cranfield, Bedford MK43 0AJ,
multi-phase f l o w The Hague, The Netherlands UK
32nd A S M E International gas 31 May-4 June 1987 IGTC, 4250 Perimeter Park South, # 108 Atlanta, Georgia, 30341,
turbine conference and exhibit Anaheim, California, USA USA
Twentieth iidwestern 31 August-2 September 1987 Professor Hamilton, School of Mechanical Engineering, Purdue
mechanics conference (20th Indiana, USA University, West Lafayette, Indiana 47907, USA
MMC)
1987 A S M E cogen-turbo 2-4 September 1987 IGTC, 4250 Perimeter Park South, # 108 Atlanta, Georgia 3041,
Montreux, Switzerland USA
6th turbulent shear f l o w 7-9 September 1987 Professor F. W. Schmidt, Secretary, Dept of Mechanical
symposium Toulouse, F,ance Engineering, The Pennsylvania State Univ., University Park, PA
16802, USA
Second international 21-23 September 1987 Dr S. M. Fraser, LDA Conference, Mechanical and Offshore
conference on laser Glasgow, Scotland, UK Engineering, University of Strathclyde, Glasgow G1 1XJ, UK
anemometry
International symposium on 15-20 November 1987 Dr J. H. Kim, Electric Power Research Institute, PO Box 10412,
natural circulation New York, NY, USA Palo Alta, California 94303, USA
1988
International fluid p o w e r 18-22 April 1988, Mr Mike Wells, International Fluid Power, Exhibition, Station
exhibition Birmingham, UK Approach, Long Lane, HiUingdon, Middlesex UB10 9NR, UK

318 V o l 7, N o 4. D e c e m b e r 1 9 8 6