ISSN 0020-4412, Instruments and Experimental Techniques, 2022, Vol. 65, No. 1, pp. 162–174. © Pleiades Publishing, Ltd.

, 2022.

LABORATORY
TECHNIQUES

Measurement Method of Nanofluids Average Velocity

Based on Laser Speckle Image
Jiajie Leia, Meiling Lianga, Shan Qinga,*, Zhumei Luoa,**, and Aimin Zhanga
a State Key Laboratory of Complex Nonferrous Metal Resources Clean Utilization, Department of Metallurgical and Energy
Engineering, Kunming University of Science and Technology, Kunming, Yunnan, 650093 China
*e-mail: [email protected]
**e-mail: [email protected]
Received July 12, 2020; revised February 2, 2021; accepted July 26, 2021

Abstract⎯In this paper, based on the optical properties of speckle, the correlation between laser speckle
image and nanoparticles is analyzed, and a method to measure the flow velocity of nanofluids using speckle
image is proposed. The influence of nanofluids type, temperature and concentration on speckle image was
investigated. On the basis of grating spatial filtering, based on PIV (Particle Image Velocimetry) cross-cor-
relation algorithm and optical flow field algorithm, the principle of gray conservation and interpretation win-
dow are introduced to extract the spectral characteristic curve of speckle image signal. Considering the inter-
ference of the environment noise, the wavelet packet decomposition method is proposed to separate the low-
frequency features and obtain the peak frequency of the signal. The static and dynamic speckle images are
analyzed by establishing a circular tube flow measurement system with laser speckle circulation to verify the
rationality and feasibility of the velocity measurement method. The experimental results show that the speckle
image of CuO nanofluids in static experiment is better than that of Al2O3 nanofluids, and increasing tempera-
ture has no obvious effect of improving image quality. The average relative error between the calculated results
and the measured results is 4.9 and 4.5%, which proves that this method is reasonable and feasible.

DOI: 10.1134/S0020441222010055

1. INTRODUCTION ence measurement of the convection field, and has

At present, with the development of science, tech- certain significance in improving the measurement
nology and industry, the research direction of nano- accuracy [5, 6]. In the non-contact method, particle
fluids is mainly focused on the heat transfer capacity image velocimetry (PIV) has the advantages of full
[1–3]. It is found that the dynamic characteristics of field, multi-point and high resolution, which is the
nanoparticles have far-reaching significance to explore most widely used measurement method in experimen-
the thermal conductivity mechanism of nanofluids and tal fluid mechanics at present [7–9]. Han L. [10],
improve the thermal conductivity of nanofluids. The improves the PIV algorithm by Kriging interpolation.
heat transfer capacity of nanofluids is closely related to The experiment shows that the improved algorithm
the distribution of nanoparticles and their Brownian has a maximum measurement error of 5.9%. And the
motion in the base solution [4]. Therefore, some peo- algorithm can compensate for the local missing flow
ple began to study the distribution and stability of field and improve the accuracy of velocity measure-
nanoparticles, and some people studied the Brownian ment.
motion of nanoparticles in the base solution. Particle image velocimetry is mainly used in optical
In the preparation of nanofluids, scanning electron metrology technology. When laser irradiates the tracer
microscopy is currently used to study the distribution particles in the fluid, a large number of obvious spots
and stability of nanoparticles in the fluid. However, are generated, which can well reflect the flow state of
how to characterize the distribution of nanoparticles the fluid. These spots are called speckle [11–13].
in the flow process is still under study. At present, the Through the processing and analysis of the speckle, the
measurement methods of flow information are mainly effective information about the flow state of the fluid
contact fluid measurement and non-contact fluid can be extracted [14, 15]. Ming Qian [16], studied the
measurement. The main advantage of contact mea- relationship between the movement of nanoparticles
surement technology is its high accuracy, but it also and speckle in laminar nanofluids. The results show
has the disadvantages of limited measurement range, that the laser speckle method can reflect the move-
too strong interference on the flow field and so on. ment of nanoparticles in nanofluids. Gong J.M. [17],
The non-contact method can realize the non-interfer- uses a line array CCD camera to measure the velocity

162
 MEASUREMENT METHOD OF NANOFLUIDS AVERAGE VELOCITY 163

(а) 500 nm (b) 500 nm

Fig. 1. SEM of CuO nanoparticles (a) and Al2O3 nanoparticles (b).

distribution of spherical and irregular particle flows. image. The steps of specific experimental are as fol-
The results demonstrate the practicality of the method lows.
in particle systems and give optimal thresholds for the 1. In the static speckle experiment, the nanofluids
system parameters. is filled into the tube, and the relative position of the
In this paper, an improved spatial filtering model of laser is adjusted to make the screen image clear. Then,
simulated grating is proposed. The spectral character- static speckle sequence images of deionized water,
istic curve of speckle image is extracted by the princi- CuO and Al2O3 nanofluids with different concentra-
ple of gray conservation and the method of interpreta- tions and temperatures were taken respectively. The
tion window. Wavelet packet [18–20] decomposition static speckle image is analyzed to determine the con-
is used to reduce noise interference in the analysis and ditions of dynamic experimental.
processing of laser speckle image of nanofluids. The 2. According to the determined conditions,
concentration of nanofluids was determined by static dynamic speckle experiment is set up. Turn on the cir-
speckle experiment, and the feasibility of the method culating water pump, and use the high-speed camera
was verified by dynamic speckle experiment. to take the dynamic speckle image when the nanoflu-
ids flows. Then, the velocity of nanofluids is estimated
based on the principles of spatial filter and wavelet
2. MATERIALS AND EXPERIMENTAL SYSTEM packet transformation, and compared with the exper-
2.1. Preparation of Nanofluids imental data obtained by the flowmeter to verify the
feasibility of the method.
In this experiment, CuO and Al2O3 nanofluids with
different mass concentration were prepared by the The device of specific experimental is shown in
two-step method with water as base solution. In order Fig. 2.
to avoid agglomeration, sodium dodecyl benzene sul- In the whole experimental platform, the main
fonate was added as dispersant, and ultrasonic vibra- equipment includes: high-speed camera, semicon-
tion was carried out for 90 min. The concentration ductor laser (wavelength 532 nm, power 100 mW, laser
range of the preparation mainly includes: 0.001, 0.005, diameter 2 mm), black light screen, spherical regulat-
0.01, 0.05, 0.1 wt%. The prepared nanofluids were char- ing valve, rotameter (40−400 L/h), circulating water
acterized by scanning electron microscope (SEM). The pump (max. 20 L/min), circular pipeline (considering
results are shown in Fig. 1. the light transmission, quartz round glass pipe with
inner diameter of 13 mm is used in the measurement
The results of SEM analysis show that the average area, and the rubber pipe is used for the rest of part).
particle size of CuO nanoparticles is 20 nm, the spe-
cific surface area is 13.1 m2/g, and the structure is
spherical. The average particle size of Al2O3 nanopar- 2.3. Experimental Research Methods
ticles is 20 nm, and the specific surface area is The research method of this paper includes two
150−200 m2/g. The two kinds of nanoparticles have parts: experimental research and numerical simula-
good dispersion and no agglomeration phenomenon, tion. In the experimental study, the speckle image is
which meets the requirements of this experiment. obtained through the experimental platform. In our
experiments, the size of the nanoparticles 20 nm was
2.2. Experimental System and Process much smaller than the wavelength of the irradiating
light 532 nm, and therefore the case of Rayleigh scat-
First, the static speckle experiment is carried out to tering is realized. In the numerical analysis, the image
analyze the experimental factors, then the dynamic is processed by MATLAB 2014a, and the nanofluids
speckle experiment is carried out according to the velocity is calculated by the improved grating spatial
experimental factors to obtain the dynamic speckle filtering method.

INSTRUMENTS AND EXPERIMENTAL TECHNIQUES Vol. 65 No. 1 2022

164 LEI et al.

Regulating valve Flowmeter

Computer
Laser

b e
tu
High speed camera

ss
gla
nt
are
sp
an
Tr
Optical screen

Pump
Liquid storage tank

Fig. 2. Schematic diagram of circular pipe channel flow measurement system.

3. SPECKLE IMAGE ANALYSIS METHOD cross-correlation algorithm and optical flow algo-
3.1. Improved Grating Spatial Filtering Method rithm. To avoid exposure area, the cross-correlation
for Velocity Measurement algorithm is used to divide the sensitive area which can
extract motion information of nanoparticles into win-
3.1.1. Basic principle of traditional grating spatial dows, and the sensitive area in the window is sampled
filtering velocity measurement. The principle of tradi- continuously according to grating filtering method.
tional raster spatial filter method [21–24] is to treat the The detailed partition diagram is shown in Fig. 3.
raster as a spatial filter, filter the light signal reflected
by the moving object, and then receive the filtered sig- As nanoparticles are illuminated by laser during the
nal by the photodetector. Finally, the received signal is flow process, the gray scale of the rough surface will
analyzed to get the velocity of object. show a certain distribution. Therefore, using the con-
cept of gray scale conservation in optical flow method,
Assuming that the spatial distribution function of the sensitive area is transformed into gray value matrix.
the moving particle is f(x,t), the spatial function rep- The specific process execution diagram is shown in
resented by the middle grating is h(x), the moving par- Fig. 4.
ticle moves along the x-axis direction, and the spatial
frequency of the x-direction is α, then the output The experimental dynamic speckle image is trans-
function of the grating spatial filter is: formed into a gray matrix. The sensitive region in the
gray matrix is selected by the window function, and
L
the sensitive area is sampled row by row by using the
m(t ) =  f ( x, t)h( x)dx,
0
(1) improved grating filtering method, and the spectrum
characteristic curve of the sampled image is extracted.
where L is the spatial length of the raster spatial filter. Wavelet packet transform is used to filter out the inter-
After Fourier transform and a series of mathemati-
cal transformations, the formula becomes the rela-
tionship between v and f:
v = K βf , (2)
Central bright spot
Sensitivity area

Sensitivity area

where v is the velocity of the moving particle; K is the

inherent optical parameter of the raster spatial filter; β
is the spatial cycle period of the raster filter, which is
the fixed value in the actual analysis and measure-
ment; f is the correlation peak frequency. The velocity
of moving particles can be obtained by finding f.
3.1.2. Measurement of nanofluids by improved grat-
ing spatial filtering. Due to the particularity of
nanoparticles size, the intensity of the speckle center is
strong and useful information is masked by it. In order Interpretation window A Interpretation window B
to extract useful information, the traditional raster
spatial filter method is improved by introducing the Fig. 3. Selection of sensitivity area.

INSTRUMENTS AND EXPERIMENTAL TECHNIQUES Vol. 65 No. 1 2022

 MEASUREMENT METHOD OF NANOFLUIDS AVERAGE VELOCITY 165

Original signal
START
Wavelet function

Transform speckle sequence into gray matrix Low frequency High frequency
characteristic term interference term

Use window function to select sensitive area

Low frequency High frequency
characteristic term interference term

Line by line sampling in sensitive areas

Reconstruction of spectrum curve
by inverse wavelet transform
Reconstruction of original signal to obtain peak frequency
Fig. 5. Characteristic curve schematic diagram of wavelet
packet reconstruction.
Calculation of fluid velocity based on improved method
optimal wavelet packet basis is calculated according to
the entropy standard, and the threshold value of the
END wavelet packet coefficient is quantified. The optimal
wavelet packet basis is used to separate the signal into
the feature term of low frequency and interference
Fig. 4. Flow chart of algorithm execution. term of high frequency. The above steps are repeated
until the signal eliminates the interference term of high
frequency, close to the original trend. The details of
ference of noise and other factors, and the peak fre- the principle of reconstructing curves by the wavelet
quency of the curve is displayed. Finally, according to packet are shown in the supplementary materials.
the improved spatial filtering method, the velocity of
nanofluids under different conditions is calculated.
4. IMAGE AND ANALYSIS
3.2. Reconstruction of Spectrum Characteristic Curve 4.1. Static Speckle Sequence Image
by Wavelet Packet Transform
4.1.1. Static speckle experiment of different con-
Ideally, the spectral characteristic curve obtained centration of water-based copper oxide. The speckle
by extracting information from the speckle image images of deionized water and different concentra-
using an improved raster spatial filter has a peak fre- tions of water-based copper oxide were analyzed and
quency f. However, in practical experiments, the peak compared.
frequency of the spectrum characteristic curve is
masked by noise due to fluctuation interference in the From the Fig. 6, we can see that the experimental
flow of nanofluids and the influence of environmental result is transmission speckle, and the central field has
noise. Considering the above factors, effective noise the strongest brightness. Due to the shape of the pipe,
removal is needed to achieve the accuracy of measure- the static speckle is oblate. It can be seen that there is
ment results. Therefore, in this paper we propose to an obvious trend of brightness diffusion on both sides
use the method of wavelet packet transformation to of the spot through magnifying observation, which is
eliminate the noise in the original signal, and the spec- the sensitive area to be analyzed by the improved grat-
trum characteristic curve is divided into several levels, ing spatial filtering method.
and the signal is decomposed layer by layer. The signal With the increase of concentration, spot brightness
features are analyzed adaptively, so that the feature becomes stronger and stronger. When the concentra-
terms are retained and the interference terms are tion of nanofluids increased to 0.05 wt%, the bright-
removed. By analogy, the peak frequencies can be ness of the light spot began to decrease. When the con-
clearly identified until the curve is approximately the centration is 0.1 wt%, the spot becomes unclear, but it
original trend in the process of multilevel decomposi- does not affect further image analysis. It is difficult to
tion. analyze if the concentration continues to be diluted or
Figure 5 is a schematic diagram of the curve recon- increased. Therefore, it can be seen from the above
structed by a wavelet packet. According to the original experimental results that the laser transmission effect
signal extracted from the speckle image, the wavelet of CuO nanofluid is most suitable for dynamic speckle
function and decomposition level are determined. The experiments when the concentration is 0.001−0.1 wt%.

INSTRUMENTS AND EXPERIMENTAL TECHNIQUES Vol. 65 No. 1 2022

166 LEI et al.

Deionized water 0.001 wt% Deionized water 0.001 wt%

0.005 wt% 0.01 wt%

0.005 wt% 0.01 wt%

0.05 wt% 0.1 wt%

0.05 wt% 0.1 wt%

Fig. 6. Static pattern of water-based copper oxide nanoflu-

ids with different mass concentrations.

Fig. 7. Static pattern of water-based alumina nanofluids

4.1.2. Static speckle experiment of different con- with different mass concentrations.
centration of water-based alumina. The static speckle
pattern of deionized with different concentrations of
Al2O3 nanofluid are shown in Fig. 7. nificant difference in brightness and clarity between
When the concentration of the water-based alu- the two groups of images. It indicates that increasing
mina nanofluids is 0.05 wt%, the intensity of the fluid temperature intensifies Brownian motion in the
bright spot in the center of the speckle image can be interior, but is not visually significant for improving
seen to be weakened. When the concentration is 0.1 wt%, image quality.
there is almost no bright spot in the center of the
speckle image. This is because the density of Al2O3
4.2. Dynamic Speckle Sequence Image Acquisition
nanoparticles is small, and the particles are distributed
more in the base liquid under the condition of the From the static speckle experiment, the suitable
same mass concentration. Therefore, when the laser concentration of nanofluids is 0.001−0.1 wt%. In the
irradiates the mixed suspension, the nanoparticles dynamic experiment, due to the huge experimental
absorb the laser, which makes the laser beam unable to data, in order to balance the data distribution, the
transmit the fluid effectively, so the image on the nanofluids with concentration of 0.005 and 0.05 wt%
screen is no longer clear. are taken as the research object. In the experiment, the
Compared with the speckle images of Al2O3 and exposure time is 10 ms and the image resolution is
CuO, the transmittance of Al2O3 nanofluids to laser in 1920 × 1080. In a complete period, 3049 speckle images
the same concentration range is not as good as that of are obtained to form a sequential speckle image. At the
CuO nanofluids. Therefore, in order to obtain the same time, different flow (50−400 L/h) control groups
dynamic laser speckle of nanofluids, the use of CuO were set up to improve the accuracy of measurement.
nanofluids has a better help to the improvement of When the concentration is 0.005 wt%, the dynamic
experimental results. speckle pattern of nanofluids at different flow rates is
4.1.3. Static speckle experiment of nanofluids at dif- as follows.
ferent temperatures. Because the nanoparticles move- It can be seen from Fig. 9 that with the increase of
ment in nanofluids is affected by Brownian motion, the flow rate, the exposure of the central area
which is temperature dependent. Therefore, the influ- increases, and the sensitive area gradually blurs. This is
ence of temperature on image quality is studied. Laser because the laser irradiates the fluid, and there are
speckle contrast of nanofluids at different tempera- scattered particles in the fluid. If the flow speed of the
tures is shown in Fig. 8. fluid is too fast, and the frame rate is limited when
Compare the above nanofluids speckle patterns at shooting, then camera of high speed can not capture
different temperatures. It is found that there is no sig- the moving state of particles in fast motion.

INSTRUMENTS AND EXPERIMENTAL TECHNIQUES Vol. 65 No. 1 2022

 MEASUREMENT METHOD OF NANOFLUIDS AVERAGE VELOCITY 167

ratio. The larger the speckle contrast ratio is, the less
blurry the image will be. On the contrary, the smaller
the speckle contrast ratio is, the more blurry the image
will be. In practice, the faster the nanofluids move, the
faster the nanoparticles in the fluid will follow the
movement, and the smaller the speckle contrast ratio,
CuO, 0.01 wt%, 40˚C CuO, 0.01 wt%, 55˚C and finally it will become more and more close to 0.
At this time, the speckle image will appear so-called
blur, but it will not affect the analysis of later image.

4.3. Measurement Results and Analysis

of Nanofluids Velocity
The dynamic speckle pattern obtained from the
Al2O3, 0.01 wt%, 40˚C Al2O3, 0.01 wt%, 55˚C above experiments is transformed into the raw signal
diagram formed by gray value matrix by using the
Fig. 8. Static speckle diagrams of nanofluids at different improved grating spatial filtering method, and then
temperatures. the spectrum characteristic curve is formed by Fourier
transform. But because the spectrum characteristic
curve contains a lot of noise, the effective peak signal
When the concentration is 0.05 wt%, the dynamic is covered. Therefore, wavelet packet inverse trans-
speckle images of nanofluids with different flow rates form is used to reconstruct the spectrum characteristic
are as follows. curve to get the peak signal, which is the peak fre-
It can be seen from Fig. 10 that the dynamic quency f.
speckle pattern with concentration of 0.05 wt% also Figure 11 shows the speckle image processing pro-
shows the same rule. With the increase of flow rate, cess at a concentration of 0.005 wt% and a flow rate of
the brighter the central area of the speckle pattern is, 50 L/h.
resulting in blurring of the sensitive area around the
central area. In the Fig. 11a, the a graph is the raw signal dia-
gram formed by converting the speckle image into a
In conclusion, there is a certain relationship gray scale matrix. The graph in Fig. 11b is the spec-
between the blur of image and the flow rate of nano- trum characteristic curve of the original signal after
fluids in the process of obtaining dynamic speckle. Fourier transform. It can be seen that there is a lot of
This relationship can be measured by speckle contrast noise in the curve at this point, and the effective peak

50 L/h 100 L/h 150 L/h

200 L/h 250 L/h 300 L/h

350 L/h 400 L/h

Fig. 9. Dynamic speckle image of laser at different flow rates at 0.005 wt%.

INSTRUMENTS AND EXPERIMENTAL TECHNIQUES Vol. 65 No. 1 2022

168 LEI et al.

50 L/h 100 L/h 150 L/h

200 L/h 250 L/h 300 L/h

350 L/h 400 L/h

Fig. 10. Dynamic speckle image of laser at different flow rates at 0.05 wt%.

signal is masked and no useful information can be ana- Amplitude, 107

lyzed. The graph in Fig. 11c is a curve graph recon- 7.5
structed by a wavelet packet transformation. Using the The original date
wavelet packet to eliminate noise interference and 7.3 The trend
extract low-frequency characteristic signal can obvi- 7.1
ously analyze the trend of the curve. According to
amplitude under different frequencies, characteristic 6.9
peak value is found and the fluid velocity is estimated
6.7 (a) Initial signal
by spatial filtering method.
6.5
As can be seen from the above figure, the original
signal is processed to obtain the obvious spectrum 5
4 The original date
characteristic curve of peak signal, which greatly facil- The trend
itates the calculation of flow rate. The dynamic 3
speckle patterns of different flow rates with a nanoflu- 2
ids concentration of 0.005 wt% are transformed in the 1
same way, as shown in Fig. 12. 0
–1
For the peak value of the spectrum characteristic –2 (b) Fourier transform
curve obtained by wavelet packet transformation in the –3
above figures, and combined with the formula v = Kβf, 5
the calculation speed below can be obtained: The original date
4
Q, L/h 50 100 150 200 250 300 350 400 The trend
3
Calcula-
tion speed, 0.0111 0.0119 0.0140 0.0143 0.0146 0.0158 0.0185 0.0195 2
m/s 1
The K value in the formula is the optical magnifi- 0 (c) Wavelet packet
cation of the high-speed camera. In the initial calcula- –1
0 100 200 300 400 500 600
tion, K = 1. After the initial calculation, the K value is f, Hz
further corrected by fitting the relevant coefficient
with the calculation results. β is the sampling period of
the analog improved grating filter. In the study, Fig. 11. Spectrum analysis of nanofluids with flow rate of
because of the continuous sampling, the value of β is 50 L/h: (a) initial signal, (b) fourier transform, (c) wavelet
the size of a single pixel of the high-speed camera. packet.

INSTRUMENTS AND EXPERIMENTAL TECHNIQUES Vol. 65 No. 1 2022

 MEASUREMENT METHOD OF NANOFLUIDS AVERAGE VELOCITY 169

3
Q = 150 L/h The original date Q = 200 L/h The original date
2
The trend 2 The trend

1 1
0
0
–1
–1 –2
2 2
Q = 250 L/h The original date Q = 300 L/h The original date
Amplitude, 107

The trend The trend

1 1

0 0

–1 –1

Q = 350 L/h The original date Q = 400 L/h The original date
2 The trend
2 The trend

1 1

0 0

–1 –1
0 100 200 300 400 500 600 0 100 200 300 400 500 600
f, Hz f, Hz

Fig. 12. Spectrum analysis of 0.005 wt% nanofluids under different flow conditions.

When nanofluids flow in a horizontal pipe, the results are shown in Fig. 13, which can be expressed as
cross-sectional area of the pipe is definite, and there is formula:
a positive correlation between the flow rate and veloc-
ity. The cross-sectional area is the proportional coeffi- v = K β f = 0.000023073Q. (3)
cient of the positive correlation. The above calculated It can be seen from the fitting diagram of the rela-
data are fitting with experimental results. The fitting tionship between the flow rate and velocity in the fig-
ure above that the fitting results of calculation and
experiment are ideal. The same transformation was
Nanofluids velocity, m/s performed for the dynamic speckle image of nanoflu-
0.020 ids with concentration of 0.05 wt%, as shown in Fig. 14.
The the peak value of the signal is extracted com-
0.018 bine the formula to calculate the fluid speed, as shown
below:
0.016 Q, L/h 50 100 150 200 250 300 350 400
Calculation
0.0059 0.0083 0.0103 0.0109 0.0110 0.0112 0.0128 0.0175
0.014 speed, m/s

The above calculated data are fitting with experi-

0.012 mental results. The fitting results are shown in Fig. 15,
which can be expressed as formula:
0.010
v = K β f = 0.000025329Q. (4)
0 50 100 150 200 250 300 350 400
Q, L/h From the above two concentrations of water-based
copper oxide nanofluids and the calculation of the rel-
Fig. 13. Fitting result chart of flow rate to 0.005 wt% nano- evant formula, we can see that the fitting results are
fluids velocity. ideal. By averaging the fitting results with formula (3)

INSTRUMENTS AND EXPERIMENTAL TECHNIQUES Vol. 65 No. 1 2022

170 LEI et al.

3 2.0
Q = 50 L/h The original date Q = 100 L/h The original date
The trend 1.5 The trend
2
1.0
1 0.5
0
0
–0.5
–1 –1.0
1.5
Q = 150 L/h The original date 1.5 Q = 200 L/h The original date
1.0 The trend The trend
1.0
0.5 0.5
0 0
Amplitude, 107

–0.5 –0.5
–1.0 –1.0
2.0
1.5 Q = 250 L/h The original date Q = 300 L/h The original date
The trend 1.5 The trend
1.0 1.0
0.5 0.5
0 0
–0.5 –0.5
–1.0 –1.0
2.0
1.5 Q = 350 L/h The original date Q = 400 L/h The original date
The trend 1.5 The trend
1.0 1.0
0.5 0.5
0 0
–0.5 –0.5
–1.0 –1.0
0 100 200 300 400 500 600 0 100 200 300 400 500 600
f, Hz f, Hz

Fig. 14. Spectrum analysis of 0.05 wt% nanofluids under different flow conditions.

and formula (4), a new relationship between the flow value can be substituted into formula (6). The calcula-
rate and nanofluids velocity can be obtained: tion speed of secondary calculation through the image
analysis in the experiment is shown in Tables 1 and 2.
v = K β f = 0.000024201Q. (5)
From the measurement of the average velocity of
In this study, there is a certain deviation between the nanofluids with concentration of 0.005 and 0.05 wt%
speed of nanofluids calculated by the image analysis in Tables 1 and 2, it can be seen that the maximum rel-
and the real speed recorded by flowmeter. At this time, ative error between the conversion velocity of flowme-
it is necessary to find a way to correct the deviation ter and the calculated velocity of the model is 13.4 and
from the relationship between the real flow rate and 13.6%, respectively. This is because when the velocity
the flow rate of nanofluids in the tube. Finally, we can of nanofluids flowing in the pipeline is low, the prob-
get the following relations: lem of nanoparticles settling is more serious, resulting
v = KK cβ f = 0.0020928Q. (6) in the quality of the speckle image, which affects the
accuracy of the measurement speed. However, when
By comparing formula (5) and formula (6), it can the flow velocity of nanofluids in the pipeline is large,
be concluded that the actual optical magnification the experimental equipment will vibrate due to the
coefficient of high-speed camera is Kc = 86.48, and its excessive flow velocity, resulting in the speckle image

INSTRUMENTS AND EXPERIMENTAL TECHNIQUES Vol. 65 No. 1 2022

 MEASUREMENT METHOD OF NANOFLUIDS AVERAGE VELOCITY 171

Nanofluids velocity, m/s temperatures is calculated when the concentration is

0.018 0.01 wt% and the flow rate is 200 L/h. As shown in
Table 3.
0.016 It can be obtained from the average speed experi-
mental calculation results in Table 3. With the increase
0.014
of the temperature of the nanofluids, the relative error
0.012 between the velocity measured by the flowmeter and
that measured by the model does not decrease gradu-
0.010 ally, which corresponds to the temperature speckle
pattern display. In theory, the increase of temperature
0.008 will aggravate the Brownian motion of nanoparticles,
which makes the distribution of nanoparticles more
0.006 uniform in the base liquid, which is helpful to the
0 50 100 150 200 250 300 350 400 450
transmission of laser, improving the image quality, and
Q, L/h reducing the relative error between the theoretical and
experimental results. However, the experimental results
show that increasing the temperature of nanofluids is
Fig. 15. Fitting result chart of flow rate to 0.05 wt% nano-
fluids velocity. beneficial to the Brownian motion of nanoparticles in
the fluid, but it has no significant effect on the mea-
surement of nanofluids flow velocity by the image
obtained mixed with many noises caused by environ- analysis.
mental and operational problems, which have a cer-
tain impact on the accuracy of further measuring the
average velocity of nanofluids. Most of the relative 4.5. Analysis of Factors Influencing
errors hover between 1 and 3%. On the whole, the the Measurement Error of Nanofluids Velocity
average of relative errors can reach 4.9 and 4.5%, By calculating the average velocity of nanofluids
respectively. According to the above two sets of data, flow under different conditions, the measurement
this method of speed measurement is reasonable. error is analyzed. The main factors that cause errors
are as follows:
4.4. Measurement Results of Nanofluids (1) In the experiment, there are some measurement
at Different Temperatures errors in experimental devices such as Rotameter used
When the concentration of water-based copper for flow measurement.
oxide nanofluids is 0.01 wt%, the flow rate is 200 L/h. (2) Small bubbles are likely to occur during the
Based on the analysis of the spectrum characteristic flow of nanofluids, which will affect the quality of the
curve (Fig. 16), the relative errors of the measurement obtained speckle image to a certain extent.
under the working conditions of 20, 40, and 55°C are
compared. Explore whether temperature has a greater (3) During the whole cycle, nanoparticles will stick
impact on speed measurement. and settle due to Brownian motion, which will change
According to the peak frequency of the signal in the the quality of speckle image and cause analysis errors.
spectrum characteristic curve, the average velocity of (4) Explore the influence of temperature in the
the water-based copper oxide nanofluids at different experiment. With the increase of temperature, the

Table 1. Calculation results of average velocity of 0.005 wt% water-based copper oxide nanofluids
Nanofluids Flowmeter Model calculation Relative error
flow Q, L/h speed v1 , m/s speed v2 , m/s of calculation results, %
50 0.1047 0.0963 8.0
100 0.2093 0.1813 13.4
150 0.3140 0.2935 6.5
200 0.4186 0.4064 2.9
250 0.5233 0.5364 2.5
300 0.6279 0.6397 1.9
350 0.7326 0.7488 2.2
400 0.8372 0.8514 1.7

INSTRUMENTS AND EXPERIMENTAL TECHNIQUES Vol. 65 No. 1 2022

172 LEI et al.

Table 2. Calculation results of average velocity of 0.05 wt% water-based copper oxide nanofluids
Nanofluids Flowmeter Model calculation Relative error
flow Q, L/h speed v1 , m/s speed v2 , m/s of calculation results, %

50 0.1047 0.1113 6.3

100 0.2093 0.1921 8.2
150 0.3140 0.3094 1.4
200 0.4186 0.4243 1.4
250 0.5233 0.5355 2.3
300 0.6279 0.6171 1.7
350 0.7326 0.7409 1.1
400 0.8372 0.9514 13.6

nanofluids will evaporate and the concentration will ysis, the stratification of speed is neglected, which
change, which will affect the clarity of the image. results in the error of speed measurement.
(5) The velocity of nanofluids is not uniform
during the flow process. The velocity near the wall of
the glass tube is close to zero, and the maximum veloc- 5. CONCLUSIONS
ity is at the center of the glass tube. In the actual anal- In this paper, the method of dividing the reading
window in PIV is applied to the selection of the sensi-
tive area of laser speckle image, and the obtained
Amplitude, 107 sequence speckle image is transformed into a gray
4 value matrix by the gray conservation principle in opti-
The original date cal flow method. Combining the above two methods
3 Temperature is 20˚C The trend
with traditional spatial filtering methods, an improved
2 spatial filtering model for analogue raster is presented.
1 Considering the noise interference during the experi-
0
ment, a new method of reconstructing the spectrum
characteristic curve by using the wavelet packet trans-
–1 formation is proposed to remove the noise from the
–2 original spectrum characteristic curve. Finally, noise
interference can be filtered out, and the peak value of
2.0 the signal can be retained, so as to make the measure-
The original date
1.5 Temperature is 40˚C
The trend ment result accurate. The velocity calculated by the
1.0 principle is compared with the measurement data of
the flowmeter in the experiment. The results show that
0.5 only two groups of data have relative errors of 13.6 and
0 13.4%, and the rest are basically distributed between
–0.5 1–3 and 6–9%, which proves that the model and cal-
culation method are feasible. The influence of tem-
–1.0
perature on the calculation method is discussed. The
5 The original date research shows that increasing temperature can
Temperature is 55˚C increase the Brownian motion of fluid, but it has no
4 The trend
obvious effect on the improvement of image quality.
3 By exploring the influence of nanofluids on measure-
2 ment, we found that the speckle image of CuO nano-
1 fluids is better than that of Al2O3 nanofluids.
0
–1 SUPPLEMENTARY MATERIALS
0 100 200 300 400 500 600 WAVELET PACKET TRANSFORM DETAILS
f, Hz
Firstly, the coefficients of orthogonal low-pass fil-
Fig. 16. Spectrum characteristic curve of nanofluids at dif- ter and high pass filter are defined as ls and hs, respec-
ferent temperatures. tively, and the following conditions must be met:

INSTRUMENTS AND EXPERIMENTAL TECHNIQUES Vol. 65 No. 1 2022

 MEASUREMENT METHOD OF NANOFLUIDS AVERAGE VELOCITY 173

Table 3. Calculation results of average velocity of 0.01 wt% CuO nanofluids at different temperatures
Nanofluids Flowmeter Model calculation
Relative error, %
temperature, °C speed v1, m/s speed v2, m/s
20 0.4186 0.4301 2.8
40 0.4186 0.3913 6.5
55 0.4186 0.4000 4.4

 ls −2ls −2k = δ j,k ,

hk = (−1) h1− k .
k
 ls = 2,
(1)
i
=  12   e
k
  t
 T0 − i
 2
0,0,k lk − 2i ( )
(8)
If the sequence of wavelet packets is defined as:
{Ts (t )|s ∈ N1} , then there are:
+
i
1  e l T t − i .


2 k

0,0,k k − 2i  0
 2
( )
T2 s (t ) = 2  lkTs (2t − k ), In the above formula, let e1,0,k be:
 (2)
T2 s +1(t ) = 2  hkTs (2t − k ). t

The scaling function of wavelet packet function is: e1,0,k = e

g
0,0, g l g − 2k , e1,1,k = eg
0,0, g hg − 2k (9)

∏ H  ω  .
T0(t ) = F
−1
The formula (9) is further substituted into (8) to
 0  j  (3)
 2  simplify:
Where the frequency response of ls is H0(w), as fol-
lows: f (t ) =  12 e
k
1,0,kT0 (2t − k ) +  12 e (2t − k ) (10)
k
1,1,k

H 0 (ω) = 1
2 k
lk e − jkω.  (4) Substitute the following formula into formula (8):
Then, from formula (2), we can get:
( )
T0 t − i , ( )
T1 t − i . (11)

( )
2 2
Ts (t − k ) = 1
2 i

lk −2iT2 s t − i
2 There are:

( )
(5)
 2 −1
J
+ 1 t
hk −2iT2 s +1 − i .
2 i 2 f (t ) = w
s = 0 k∈Z
J , s, k 2
− J /2
Ts (2
−J
− k ); (12)
It can be seen from the above formula that on the
left side of the formula, it constitutes a set of normal
orthogonal bases of L2(R).  m

w j,2 s,k = w j −1,s, g l g −2k ,
 (13)
The continuous time signal of discrete sampling
signal is given as follows:
 j,2 s +1,k

w =
m
w  h
j −1, s, g g − 2k .

The above algorithm is called tower algorithm, in

f (t ) = ek
0,0,kT0 (t − k) (6) which j, s can be expressed as the following formula:
where e0.0,k represents the k-th component of the dis- j = 1,2,3,… , J ,
(14)
crete sampling signal. Formula (6) is similar to Shan- s = 0,1,… ,2 − 1.
j
non’s theorem. Then from formula (5), we can get:
Then, the corresponding formula is as follows:
T0(t − k ) = 1 
2 t 2 ( )
lk −2iT0 t − i + 1
2 t 2

hk −2iTt t − i . (7) ( ) w j −1,s,k = l
g
k − 2 g w j,2 s, g + h s
k − 2 g w j,2 s +1. g . (15)

Formula (7) is combined with formula (6), and the From the above deduction, we can see that formula
dissolving formula (6) is as follows: (13) is the wavelet packet decomposition algorithm, and

)  l T (2t − i )
formula (15) is the wavelet packet recovery method.
1
f (t ) = e
k
0,0,k 
 2 i
k − 2i 0
CONFLICTS OF INTEREST
T ( t − i )

+ 1
2
hi
k − 2i 1
2 
The authors of this work declare that there are no con-
flicts of interest regarding the publication of this paper.

INSTRUMENTS AND EXPERIMENTAL TECHNIQUES Vol. 65 No. 1 2022

174 LEI et al.

ACKNOWLEDGMENTS 11. Adrian, R.J., Exp. Fluids, 2005, vol. 39, no. 2, p. 159.
This work was supported by National Natural Science https://doi.org/10.1007/s00348-005-0991-7
Foundation of China under Contract (nos. 51966005, 12. Park, J.H., Choi, W., and Sang, J.L., Ultrasonics, 2020,
52069010). vol. 104, p. 106093.
https://doi.org/10.1016/j.ultras.2020.106093
13. Contado, E., Júnior, R.A.B., Barbosa, H.C., Reis, R.O.,
REFERENCES and Prabhu, R., Theor. Appl. Eng., 2020, vol. 5, p. 1.
1. Motevasel, M., Solaimany Nazar, A.R., and Jamialah- https://doi.org/10.31422/taae.v4i4.33
madi, M., J. Therm. Anal. Calorim., 2019, vol. 135, 14. Adrian, R.J. and Lyilo, C.S., Appl. Opt., 1985, vol. 24,
p. 133. p. 44.
https://doi.org/10.1007/s10973-018-7314-8 https://doi.org/10.1364/AO.24.000044
2. Eastman, J.A., Choi, U.-S., Li, S., Thompson, L.J., 15. Peisen, L., Basis of Speckle Statistical Optics, Science
and Lee, S., MRS Proc., 1996, vol. 457, p. 3. Press, 1987, pp. 5−56, pp. 91−103.
https://doi.org/10.1557/PROC-457-3
3. Wang, X. and Xu, X., J. Thermophys. Heat Transfer, 16. Ming Qian, Xiao Wu Ni, Jian Lu, and Zhong Hua
2012, vol. 13, no. 13, p. 474. Shen, Key Eng. Mater., 2008, vol. 86, p. 1111.
https://doi.org/10.2514/2.6486 https://doi.org/10.4028/www.scientific.net/KEM.364-366.
1111
4. Eastman, J.A., Yu, W., Choi, U.-S., Thompson, L.J.,
and Li, S., Appl. Phys. Lett., 2001, vol. 78, no. 6, p. 718. 17. Gong, J.M., Yang, H., Lin, S.H., Li, R., and Zivkovic, V.,
https://doi.org/10.1063/1.1341218 Powder Technol., 2018, vol. 324, no. 15, p. 76.
5. Alimenti, F., Bonafoni, S., Gallo, E., Palazzi, V., Gatti, R.V., https://doi.org/10.1016/j.powtec.2017.10.041
Mezzanotte, P., Roselli, L., Zito, D., Barbetta, S., and 18. Zhang, D. and Liu, R., Int. J. Opt., 2019, vol. 2019.
Corradini, C., IEEE Trans. Geosci. Remote Sens., 2020, no. 9, p. 1.
vol. 58, no. 7, p. 5195. https://doi.org/10.1155/2019/1097292
https://doi.org/10.1109/TGRS.2020.2974185 19. Sun, K., Zhang, J., Shi, W., and Guo, J., Energies,
6. Kazuya Nakano, Yuta Aoki, Ryota Satoh, Akira Hoshi, 2019, vol. 12, no. 17, p. 3242.
Hiroyuki Suzuki, and Izumi Nishidate, Proc. SPIE https://doi.org/10.3390/en12173242
9715, Optical Diagnostics and Sensing XVI: Toward
Point-of-Care Diagnostics, San Francisco, CA, 2016, 20. Bettayeb, F., Haciane, S., and Aoudia, S., NDT&E Int.,
p. 971515. 2005, vol. 38, no. 6, p. 478.
https://doi.org/10.1117/12.2212417 https://doi.org/10.1016/j.ndteint.2004.12.003
7. Zhang Jun, Chen Liujun, Xu Qi, Wu Yungang, and 21. Feng Zhen and Ma Qishuang, Laser Infrared, 2010,
Zhang Long, J. Aeronaut. Power, 2017, vol. 32, no. 06, vol. 40, no. 3, p. 315,
p. 1289. https://doi.org/10.3724/SP.J.1087.2010.02819
https://doi.org/10.13224/j.cnki.jasp.2017.06.002 22. Kan, Z., Shao, F., and Li, Q., Chin. J. Sens. Actuators,
8. Soares, R.R., Barbosa, H.C., Braga, R.A., Botega, J.V.L., 2012, vol. 25, no. 5, p. 609.
and Horgan, G.W., Flow Meas. Instrum., 2013, vol. 30, https://doi.org/10.3969/j.issn.1004-1699.2012.05.011
p. 90. 23. Gao, H., Fan, B., and Chang, Q., Rev. Sci. Instrum.,
https://doi.org/10.1016/j.flowmeasinst.2013.01.008 2019, vol. 90, no. 7, p. 075002.
9. Masayasu Shimura and Mamoru Tanahashi, J. Com- https://doi.org/10.1063/1.5100184
bust. Soc. Jpn., 2019, vol. 61, no. 197, p. 231.
https://doi.org/10.20619/jcombsj.61.197_231 24. Xu, C., Zhou, B., Yang, D., Tang, G., and Wang, S.,
Meas. Sci. Technol., 2008, vol. 19, no. 2, p. 211.
10. Han, L., Wang, H., Liu, X., Xie, R., and Fu, C., Sen-
https://doi.org/10.1088/0957-0233/19/2/024005
sors, 2019, vol. 19, no. 12, p. 2702.
https://doi.org/10.3390/s19122702

INSTRUMENTS AND EXPERIMENTAL TECHNIQUES Vol. 65 No. 1 2022