The problem of interaction of
the two-phase (dispersed) flows with streamlined bodies
arises
as part of analysis of
the motion of various aircrafts in a
dusty or rain atmosphere [1–3], and the flowing of two-phase coolants in
channels of power facilities [4–6] as well. The presence of solid particles or
drops may cause a significant (in some instances even multiple)
increase
of
heat fluxes and erosion wearing of the streamlined surface. These phenomena are
considered as a part of one problem due to
a number of reasons: changing of
characteristics of the incoming flow [7] and parameters of the boundary layer
appearing on the streamlined body [8], appearing of vortex zones and turbulence
traces, collisions of particles (or droplets) with the surface of the body [9,
10], changing of the surface roughness etc. The intensity of the processes
that
accompanies
the two-phase flow around bodies
depends on the inertia and the particle (droplets) concentration.
The main tasks of studying of
two-phase flow around bodies are: 1) investigation of particles motion and
determination of their characteristics – velocity fields, concentration etc. 2)
determination of the influence of particles on the gas flow mode; 3)
investigation of the interaction of particles with the streamlined surface
[11], including erosion wearing, gas-dynamic sputtering [12–14], icing [15–18],
illumination [19, 20] etc.
One of the tasks of the
experimental investigation of two-phase flows is the determination of particles
concentration fields in channels and in the vicinity of streamlined bodies as
well. When performing these experiments, there is frequently a problem of
ensuring the uniform spatial particles distribution at the entrance to the
considered flow area. To estimate the uniformity of particles distribution over
the channel cross-section area, we propose an algorithm determining their
concentration after the series of frames shot.
This work is aimed to the
visualization of time-averaged fields of particles concentration in the
two-phase flow through a channel using high-speed video recording with the
subsequent algorithmic processing of images obtained.
The experimental unit to
investigate the two-phase flows (Fig. 1) is a vertical channel. The lemniscate
inlet with feeder
1
is mounted at the entrance of the channel,
through which is fed the disperse phase. The flow passes initial path
2
made
of PVC and having the diameter of 100 mm and the length of 1000 mm. Then via
transfer path
3
the flow enters operational path
4,
made of organic transparent glass (the square cross-sectional side 100 mm). The
length of operational path
4
is 500 mm. Inside operational path
4
is mounted streamlined cylindrical body with a flat horizontal surface
5
having the diameter of 17 mm. Outlet path
6
connects operational
path
4
and cyclone type dust collector
7, onto
which fan
8
with the variable flow rate is mounted. The dispersed
phase settles in the bins
9, mounted on cyclone
7
and outlet path
6. The experimental unit is mounted on steel
frame
10.
Lemniscate inlet
1,
transfer path
3, cylindrical body
5
and bins
9
are manufactured of polylactide (PLA) with the use of additive FDM-technology.
The flow rate of particles is
varied within the range of 1 - 10 g/s by the variation of the feeder
cross-section area. The fan allows obtaining
the
airflow velocities up to 8 m/s.
In
the process of video shooting,
we use the
lighting by LED floodlight
11
that is
installed perpendicular
to the channel axis (in the frame plane). To increase the uniformity of the
floodlight illumination of the working area diffusion filters are used.
Fig. 1. Experimental unit for
the two-phase flow visualization:
1
– lemniscate inlet with feeder;
2
– initial path;
3
– transfer path;
4
– operational path;
5
– streamlined body;
6
– outlet path;
7
– cyclone type dust
collector;
8
– fan;
9
– bins;
10
– frame;
11
– LED
floodlight;
12
– high-speed camera
As a dispersed phase in the
experiments are used glass microspheres (Fig. 2) (the physical density of
particles is 2550 kg/m3) with the averaged diameter of 165
μm (the mean square deviation of the diameter is 18
μm).
|
|
|
|
a
|
b
|
Fig. 2. Photo of particles at the 80x
magnification (a); histogram of the particles diameter distribution (b)
The
video shooting of the flow volume under investigation is performed by the
high-speed monochrome camera F 1500-32-Ì
(Evercam,
Russia) (Fig. 3). Its maximum resolution is 1920õ1088
pixels at a shooting speed of 1500 frames per second. The frame rate can be
increased to 44000 frames per second, but in this case the resolution drops to
1920õ16 pixels. Zenitar Zenit 1,2/50s lens (Zenit,
Russia) was used for shooting.
Fig. 3. High-speed monochrome
camera Evercam F 1500-32-Ì
The unit has the following
parameters: flow velocity (on the operational path axis) – 3 m/s; particles
mass flow rate – 2,4 g/s; air flow rate – 36,6 g/s. Video-shooting parameters:
exposition time – 1/1500 s; frame frequency – 2 Hz; lens aperture – f/2,0;
focal length of the lens – 50 mm; resolution of frames – 1920õ1088 pixels. The experiment duration was 360 seconds.
For the
problem discussed in
this
work,
we
have
developed
the
software written in the MATLAB
environment with use of the Image Processing Toolbox library.
The
initial data is a
set of
monochrome images (Fig. 4a)
that are
obtained experimentally.
For the algorithm processing, the
images
were cropped
to the region of interest and their inversion in the case of
shooting with a shadow background method. Subsequently, the tone correction of the
image was
done, to obtain a sharper image of the particle boundary.
Then
minimal value of each pixel for the total set of frames
was computed. As a result, we obtained the
image where there were no particles, but the elements of the unit were
presented (walls of operational part, object under investigation, i.e.
background image). The image of the minimum values was subtracted from the
current set, pixel-by-pixel (Fig. 4b). Particle boundaries became sharper and background
noise increased significantly.
|
|
|
|
a
|
b
|
Fig.
4.
Initial image (à);
cropped inverted image with the subtracted background (b)
The particle boundaries are
detected with
the
technique developed by J. F. Canny
[21]. The latter allows
obtaining
the
binary image of boundaries. Closed areas imaging the particles are
filled in. False images of open boundaries are discarded (Fig. 5a) using the
morphological opening operation.
From
the binary particle image,
some
parameters
are estimated, i.e.,
position of the centers of mass of the particles (Fig. 5b), area, initial
particle image, direction of the main central axes.
Each image
representation is extended with a
set of
estimated parameter values
assigned for
the found particles.
|
|
|
a
|
b
|
Fig.
5.
Particle boundary binary image (a);
particle center of mass locations (b)
We
filter the obtained set of the images by
the average value of the particle image
intensity (Fig. 6), thus considering only those that are
essential
for
the further analysis. During the shooting, one should keep in mind
the depth of the sharply depicted space: when the aperture is open, although
the aperture of the optical system is the largest, the area of sharpness of the
image is reduced. This results in, that the images of particles near the back
and front wall of the operational pass become unsharp. (Fig. 6b). In this case,
the identified particles are less likely to pass the average image intensity
test.
|
|
|
|
a
|
b
|
Fig.
6.
Examples of single particle images: sharp
image (a); unsharp image (b)
One of the key characteristics
of two-phase flows is the spatial distribution of particles. The data obtained
with the use of particle identification algorithm may be represented
with
a
so-called heat map. The area of the operational pass image to be processed is
split into cells, each of 20õ20 pixels. Subsequently,
it is calculated the number of particles in each cell over the total set of
photos (Fig. 7a). The ratio of the number of particles in a cell under
calculation
over all the frames to the total number
of frames
determines the average quantity of particles
np
in the
given cell.
The feeding of particles into
the flow occurs via the hole (4 mm of diameter) in the feeder, what is 1/25 of
the transversal size of the operational channel. From Fig. 7b one can see, that
the disperse phase, after passing the initial path, comes into the operational
path with the non-uniform distribution over the cross-section area: there is a
concentration growing from the walls to the channel axis. The opening angle of
the jet relative to the channel axis is approximately 5-7°. When the flow
approaches the end of the streamlined body, a sharp increase in concentration
takes place (in 2 and more times), due to the contribution of reflected
particles moving in the opposite direction towards the main flow. In the
conditions of the low-dust flow, the oncoming particles reach a height of about
three diameters of the end face what enables the effect of surface shielding.
|
|
|
a
|
b
|
Fig.
7.
Distribution of particles over the operational
pass: heat map of square cells (a); contour plot (b)
We have
discussed the problem of the detection of
the fields of particles concentration in a multiphase flow with the
use of the algorithm developed by J. F. Canny. Application of the technique
proposed to the flow with the glass microspheres showed a fairly reliable
result, which is consistent with the experimental images taken separately from
the sample. There had been obtained particles distributions over the
operational path cross-section area and their frame-by-frame images. The
further visual interpretation proposed demonstrates, e.g., a rapid growing of
the particle concentration in the vicinity of the streamlined body. The proposed
technique can be of interest for the investigation of surface shielding in
dusted flows.
1. Teoriya i proektirovanie gazoturbinnyh i kombinirovannyh
ustanovok [Theory and Design of Gas Turbine and Combined Plants]. Varaksin
A.Yu. Ed. Moscow:
Mosk. Gos. Tekh. Univ. im. N.E.
Baumana, 2017.
2. Mikhatulin D.S., Polezhaev Yu.V., Reviznikov D.L. Teploobmen
i razrushenie tel v sverhzvukovom geterogennom potoke [Heat Transfer and
Destruction of Bodies in a Supersonic Heterogeneous Flow]. Moscow: Yanus-K,
2007.
3. Mikhatulin D.S., Polezhaev Yu.V., Reviznikov, D.L.,
Teplomassoobmen, termokhimicheskoe i termoerozionnoe razrushenie teplovoi
zashchity [Heat and Mass Transfer, Thermochemical and Thermal Erosion
Destruction of Thermal Protection]. Moscow: Yanus-K, 2011.
4. Sukomel
A.S., Tsvetkov F.F., Kerimov R.V. Teploobmen i gidravlicheskoe soprotivlenie
pri dvizhenii gazovzvesi v trubakh [Heat Transfer and Hydraulic Resistance
during the Movement of Gas Suspension in Pipes]. Moscow: Energiya, 1977.
5. Deich M.E.,
Filippov G.A. Gazodinamika dvukhfaznykh sred [Gas Dynamics of Two-Phase Media].
Moscow: Energoizdat, 1981.
6. Perel’man
R.G., Pryakhin V.V. Eroziya elementov parovykh turbin [Erosion of Steam Turbine
Components]. Moscow: Energoatomizdat, 1986
7. Varaksin A.Yu.
Effect of
particles on carrier gas flow turbulence. High Temperature, Vol. 53, ¹ 3, 2015,
P. 423
–
444.
8. Varaksin
A.Yu.
Two-phase boundary layer of gas with solid particles. High
Temperature, Vol.58, ¹ 5, 2020, P. 716
–
732.
9. Varaksin A.Yu.
Collision of
particles and droplets in turbulent two-phase flows. High Temperature, Vol.57,
¹ 4, 2019, P. 555
–
572.
10. Varaksin A.Yu.
Two-phase
flows with solid particles, droplets, and bubbles: problems and research
results (review). High Temperature, Vol.58, ¹ 4, 2020, P. 595
–
614.
11. Varaksin
A.Yu. Collisions in Particle-Laden Gas Flows. New York: Begell House, 2013.
12. Kiselev S.P., Kiselev V.P., Klinkov S.V., Kosarev V.F.,
Zaikovskii V.N. Study of the gas-particle radial supersonic jet in the cold spraying.
Surface & Coatings Technology, Vol. 313, 2017, P. 24-30.
13. Klinkov S.V., Kosarev V.F., Shikalov V.S. Influence of
nozzle velocity and powder feed rate on the coating mass and deposition
efficiency in cold spraying. Surface & Coatings Technology, Vol. 367, 2019,
P. 231-243.
14. Fomin V.M., Golyshev A.A., Kosarev V.F., Malikov A.G.,
Orishich A.M., Filippov A.A. Deposition of cermet coatings on the basis of Ti,
Ni, WC, and B4C by cold gas dynamic spraying with subsequent laser irradiation.
Physical Mesomechanics, Vol. 23, 2020, P. 291-300.
15. Fortin G., Ilinca A., Laforte J.-L., Brandi V. A new
roughness computation method and geometric accretion model for airfoil acing.
J. of Aircraft, Vol. 41, ¹ 1, 2004, P. 119-127.
16. Alekseenko S.V., Prikhod`ko A.A. Mathematical modeling
of ice body formation on the wing airfoil surface. Fluid Dynamics, Vol. 49, ¹
6, 2014, P. 715-732.
17. Takahashi T., Fukudome K., Mamori H., Fukushima N.,
Yamamoto M. Effect of characteristic phenomena and temperature on super-cooled
large droplet icing on NACA0012 airfoil and axial fan blade. Aerospace, Vol. 7,
¹ 7, 2020, Paper No. 92.
18. Min S., Yee K. New roughness-induced transition model
for simulating ice accretion on airfoils. AIAA Journal, Vol. 59, ¹ 1, 2021, P.
250-262.
19. Molleson G.V., Stasenko A.L. Gas thermodynamics and
optics of a monodisperse supersonic jet interacting with an aerodynamic body.
High Temperature, Vol. 50, ¹ 6, 2012, P. 755-764.
20. Molleson G.V., Stasenko A.L. Electro-optical phenomena
in a gas-dispersed jet flow around a solid body. High Temperature, Vol. 53, ¹
6, 2015, P. 855-864.
21. Canny J. F. A computational approach to edge detection.
IEEE Trans. Pattern Analysis and Machine Intelligence, Vol. 8, 1986, P.678-698.
