Group Title: 7th International Conference on Multiphase Flow - ICMF 2010 Proceedings
Title: 11.6.4 - Properties of primary and secondary waves in annular gas-liquid flow
ALL VOLUMES CITATION THUMBNAILS PAGE IMAGE ZOOMABLE
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/UF00102023/00293
 Material Information
Title: 11.6.4 - Properties of primary and secondary waves in annular gas-liquid flow Multiphase Flows with Heat and Mass Transfer
Series Title: 7th International Conference on Multiphase Flow - ICMF 2010 Proceedings
Physical Description: Conference Papers
Creator: Alekseenko, S.V.
Cherdantsev, A.V.
Heinz, O.M.
Kharlamov, S.M.
Markovich, D.M.
Publisher: International Conference on Multiphase Flow (ICMF)
Publication Date: June 4, 2010
 Subjects
Subject: annular flow
primary waves
secondary waves
secondary instability
high-speed LIF technique
 Notes
Abstract: Wavy structure of annular gas-liquid flow is studied using high-speed laser-induced fluorescence technique. Automatic algorithm is developed for processing spatio-temporal records of local film thickness. Flow regimes without liquid entrainment are investigated in order to obtain characteristics of primary and secondary waves, such as velocity and amplitude, length of front and rear slopes, and relative coordinates of points of generation of secondary waves.
General Note: The International Conference on Multiphase Flow (ICMF) first was held in Tsukuba, Japan in 1991 and the second ICMF took place in Kyoto, Japan in 1995. During this conference, it was decided to establish an International Governing Board which oversees the major aspects of the conference and makes decisions about future conference locations. Due to the great importance of the field, it was furthermore decided to hold the conference every three years successively in Asia including Australia, Europe including Africa, Russia and the Near East and America. Hence, ICMF 1998 was held in Lyon, France, ICMF 2001 in New Orleans, USA, ICMF 2004 in Yokohama, Japan, and ICMF 2007 in Leipzig, Germany. ICMF-2010 is devoted to all aspects of Multiphase Flow. Researchers from all over the world gathered in order to introduce their recent advances in the field and thereby promote the exchange of new ideas, results and techniques. The conference is a key event in Multiphase Flow and supports the advancement of science in this very important field. The major research topics relevant for the conference are as follows: Bio-Fluid Dynamics; Boiling; Bubbly Flows; Cavitation; Colloidal and Suspension Dynamics; Collision, Agglomeration and Breakup; Computational Techniques for Multiphase Flows; Droplet Flows; Environmental and Geophysical Flows; Experimental Methods for Multiphase Flows; Fluidized and Circulating Fluidized Beds; Fluid Structure Interactions; Granular Media; Industrial Applications; Instabilities; Interfacial Flows; Micro and Nano-Scale Multiphase Flows; Microgravity in Two-Phase Flow; Multiphase Flows with Heat and Mass Transfer; Non-Newtonian Multiphase Flows; Particle-Laden Flows; Particle, Bubble and Drop Dynamics; Reactive Multiphase Flows
 Record Information
Bibliographic ID: UF00102023
Volume ID: VID00293
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: 1164-Alekseenko-ICMF2010.pdf

Full Text

7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010


Properties of primary and secondary waves in annular gas-liquid flow


S.V. Alekseenkol2, A.V. Cherdantsev'2, O.M. Heinz S.M. Kharlamov' and D.M. Markovich'2

'Institute of Thermophysics, 1 Lavrentiev ave., Novosibirsk 630090, Russian Federation
2Novosibirsk State University, 2 Pirogov str., Novosibirsk 630090, Russian Federation

E-mail: cherdantsev@itp.nsc.ru


Keywords: annular flow, primary waves, secondary waves, secondary instability, high-speed LIF technique


Abstract

Wavy structure of annular gas-liquid flow is studied using high-speed laser-induced fluorescence technique. Automatic
algorithm is developed for processing spatio-temporal records of local film thickness. Flow regimes without liquid entrainment
are investigated in order to obtain characteristics of primary and secondary waves, such as velocity and amplitude, length of
front and rear slopes, and relative coordinates of points of generation of secondary waves.


Introduction

Annular gas-liquid flow represents a motion of liquid film
along channel wall, sheared by high-velocity gas stream.
Film surface in such flow is covered by complex system of
waves of different types, interacting with each other. In
particular, at high liquid flow rates, interaction of
disturbance waves and ripple waves leads to entrainment of
liquid from film surface into the core of gas stream.
According to observations of Woodmansee & Hanratty
(1969), small-scale ripple waves (or ripples), travelling on
the crests of large-scale disturbance waves, are being
disrupted by the gas shear into tiny droplets. Large number
of works was devoted to studying the properties of
disturbance waves and ripples during the last fifty years.
In particular, amplitude, velocity, frequency, spacing and
lifetime of disturbance waves were studied in Hall Taylor &
Nedderman (1968), Chu & Dukler (1975), Azzopardi (1986),
Sekoguchi & Mori (1997), Han et al. (2006), Sawant et al.
(2008), Damsohn & Prasser (2009), etc. Large values of all
mentioned characteristics of disturbance waves in
comparison to those of ripples were observed. Ripples are
less studied: their properties were investigated in several
works (e.g., Chu & Dukler 1974) independently of
disturbance waves.
Combined study of waves of different types is possible if
data on spatio-temporal evolution of film surface with high
enough spatial and temporal evolution are available. In a
few works attempts to create such a system were done. E.g.,
in works of Sekoguchi & Mori (1997) and Damsohn &
Prasser (2009) complicated measuring systems, consisting
of large number of conductivity probes, were created.
Unfortunately, spatial resolution of those systems was
comparable to the longitudinal size of ripples.
Recent observations of the authors, performed with
high-speed laser-induced fluorescence technique (see
Alekseenko et al. 2008, 2009a), have shown that all the
ripples in such flow appear at the rear slopes of disturbance
waves. After inception, a newborn ripple may move either


faster or slower than parent disturbance wave. The relative
direction of its further travel depends on relative coordinate
of the ripple's inception.
If the ripple appears closer to the disturbance wave's crest
than certain distance, it travels faster than parent
disturbance wave, promptly reaches its front and disappears.
This disappearance of the 'fast ripples' occurs due to
scattering of ripple by the gas shear with subsequent
entrainment of droplets.
If the ripple appears farther than certain distance from crest
of disturbance wave, it initially moves with the same
velocity as parent disturbance wave does, but gradually
decelerates and slides to the residual layer ('substrate')
between disturbance waves. Far enough from parent
disturbance wave such 'slow ripple' reaches some constant
value of velocity, and normally is being absorbed by the
following disturbance wave.
If liquid Reynolds number is lower than particular value,
called critical Reynolds number, there occurs no liquid
entrainment irrespective of gas velocity. Up-to-date it was
supposed that no disturbance waves exist in such regimes,
and, thus, no ripples are travelling on disturbance waves
crests to be scattered by the gas shear. The wavy system in
such flow regimes was considered to consist of random
ripples. Such approach was used in works of Suzuki et al.
(1983), Asali & Hanratty (1993), Alekseenko et al. (2007).
Only in the work of Ohba & Nagae (1993) an attempt to
divide waves in such regimes into different types was
performed. These authors used transverse size of waves as a
criterion of separation. They concluded that, besides ripple
waves that are localized in circumference of the tube, ring
waves exist, which have transverse size equal to the tube
perimeter. Ring waves were observed in the vicinity of
transition to entrainment only, in rather narrow range of gas
velocities.
In recent works of the authors (Alekseenko et al. (2009 a,
b)) two types of waves were also observed in flow regimes
without entrainment. It was found that faster long-living
waves coexist with slower short-living waves, and the latter









always appear at the rear slopes of the former. The
short-living waves (that we called 'secondary waves') after
inception move with slower velocity than parent long-living
waves (that we called 'primary waves'), slide to residual
layer between primary waves and are being absorbed by the
following primary wave. This behaviour is very similar to
that of disturbance waves and 'slow ripples' in entrainment
condition.
Basing on this similarity, we supposed that primary and
secondary waves in regimes without entrainment belong to
the same types of waves, that, respectively, disturbance
waves and ripples in entrainment conditions belong to.
Further, for brevity and unification of terminology, we will
call disturbance waves and ripple waves in entrainment
regimes 'primary waves' and 'secondary waves',
respectively.
The explanation of transition to entrainment with growth of
liquid Reynolds number was also proposed. According to
this explanation, no entrainment occurs at low liquid
Reynolds numbers, since no secondary waves move faster
than primary ones in such conditions, and, thus, no
secondary waves can be scattered on the crests of primary
waves.
In one of the recent works (Alekseenko et al. 2009b) we
performed some simple hand-made processing of
time-space records of film thickness, which gave us the
characteristics of processes of generation of secondary
waves on the rear slopes of primary ones. In particular,
distributions of the relative coordinates of points of
generation were roughly estimated.
In this work we present new data on evolutionary
characteristics of primary and secondary waves in annular
two-phase flow, obtained using specially developed
automatic algorithm of data processing.

Nomenclature


H
Hsub
Re
Vcor
Vgas
d
lb
If
q


Film thickness (mm)
Substrate thickness (mm)
Liquid Reynolds number
Cross-correlation velocity of waves (m/s)
Superficial gas velocity (m/s)
Inner diameter of the tube (mm)
Length of back slope of primary wave (mm)
Length of front slope of primary wave (mm)
Volumetric flow rate of liquid (m3/s)


t Time (ms)
x Longitudinal distance (mm)
xc Distance from beginning of primary wave to the
point of inception of secondary wave (mm)
y Transverse distance (mm)

Greek letters
v kinematical viscosity of liquid (cSt)


Experimental Facility

Experiments were performed in vertical transparent
cylindrical channel with the inner diameter of 15 mm and
length of 1 meter, at distances 55-75 cm below the inlet.
Liquid film was formed on channel wall using slot
distributor with the gap width of 0.5 mm, gas entered the


7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010

channel through a co-axial tube with smaller diameter.
Channel was made with square outer side to decrease
optical distortions. Continuous green laser with the
wavelength 532 nm and power 50 mW was used as the light
source. Laser beam was converted into a vertical light sheet
aimed at the region of measurements. Rhodamin-6G with
the concentration 30 mg/1 was used as fluorescent dye.
Fluorescent light was registered by the 10-bit CCD camera
with linear (2,048 x 1 pixels) matrix. Camera was aimed at
the region of measurements with the focus on the nearest
inner wall of the channel. Camera was equipped with
low-pass orange filter (cut-off wavelength 550 nm), since
Rhodamin-6G emits fluorescent light with maximum in
orange spectral domain. Axes of camera and laser were
placed at a small angle to each other on horizontal plane.
This arrangement allowed avoiding registration of light
emitted by the film flowing on the distant wall of the
channel.
All experiments were conducted in the measurement area
with the length of 10 cm (60-70 cm below inlet) with spatial
resolution 0.1 mm. Width of measurements area was 0.1
mm, and the thickness of laser sheet was about 1 mm.
Exposition time was 250 ps and registration frame
rate-2,000 fps. For each set of flow parameters at least five
records of 1 second duration were made.
Registered image brightness was converted into local film
thickness using calibration curve obtained in a set of in situ
calibration tests. Calibration was made for every pixel.
During long-run experiments, there exists thermal drift of
camera but the error associated with it did not exceed 1% in
our case. The influence of liquid temperature on calibration
curve in our experimental conditions is negligible.
Liquid Reynolds numbers Re=20 and 40 were chosen for
regimes without entrainment. Re was defined as q/7dv. The
range of average gas velocities Vg was 14-52 m/s. Distilled
water was used as liquid, air was used as gas.

On 2D and 3D structure of waves

An example of spatio-temporal behaviour of local film
thickness in no-entrainment conditions is given in Figure 1.
In this Figure one can see the waves that travel across
time-space surface along characteristic lines. Velocity of a
wave is proportional to the slope of such line to time axis.
Primary waves can be observed as the waves of higher
amplitude and velocity, which generate slower and smaller
secondary waves on their rear slopes. The secondary waves
do not live for long time, since they are normally being
absorbed by the following primary wave.






7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010

S Primary wave
W direction a
Secondary waves a


Figure 1: Space-time behaviour of local film thickness
(Re=40, Vgas=44 m/s) in 2D representation.

When studying two-dimensional evolution of local film
thickness, the question always appears about how do
obtained results relate to three-dimensional structure of film
surface. In present time, we performed the first experiments
on 3D high-speed LIF visualization of annular gas-liquid
flow. High-speed camera with rectangular matrix was used
in these experiments with 2 Wt continuous green laser.
Transverse size of the investigated area was slightly higher
than 1 cm. Water-glycerol solution with v=3 cSt was used as
the working liquid. Camera frame rate was 500 fps. In
3D-experiments standard scenario of primary and secondary
waves behaviour, described in Alekseenko et al (2009a),
was observed.
In Figure 2 (a), an example of three-dimensional
LIF-visualized shape of film surface is shown. A primary
wave, travelling from right to left and taking all the
transverse size of investigated area, can be seen at the left
side of the Figure. Behind it, one can see the secondary
waves, generated by primary one. The secondary waves are
characterized by lower velocity and amplitude and they are
localized by their transverse size.
In Figure 2 (b) an example of data of Figure 2 (a) in
2D-representation (sequence of instantaneous film thickness
distributions along the line y=7 mm) is shown. It can be
concluded that 2D representation of film surface evolution
provides necessary information on processes of generation
of secondary waves, and essentially simplifies data
processing. Thus, in present stage of investigation, only the
2D-approach will be used for data processing.


Figure: Examples of shape of film surface (Re=18,
Vg=18 m/s, water-glycerol solution with v=3 cSt) in 3D
representation (a) and in 2D representation (b).

Automatic algorithm: Canny transform application

Development of automatic algorithm of data processing is
the main goal of present work. Such algorithm should
identify waves on x-t surface, separate them into primary
and secondary ones and process the two types separately,
defining their individual evolutionary characteristics. One of
the difficulties consists in fact that even the velocity of more
stable primary waves is not constant, it fluctuates with
evolution of the wave, and the algorithm should follow
these fluctuations.
Canny transform was chosen for the first stage of the
automatic processing finding the waves. This procedure is
normally used for finding edges on images. In short words,
it searches the points where response of the signal to the
Gaussian filter reaches the highest values. Details can be
found in Canny (1986). When being applied to our
time-space dependencies of local film thickness, Canny
transform marks the points with the highest slope of the
interface (further we will call such points 'Canny points').
Normally, each slope of a wave is marked by continuous
sequence of Canny points that we will call 'Canny
segments'.


H. mm





m


Flo





7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010

instead of Vcor. When the range was empty, search of new
line began. After concatenating all Canny segments,
obtained characteristic lines shorter than certain value of
lifetime were eliminated.

Automatic algorithm: waves separation and
primary waves processing

The next step was to separate the obtained Canny lines into
primary and secondary waves. Criterion of separation based
on lifetime of the waves was used. Lifetime distribution of
all waves for the same regime is given in Fig. 4 (a).


Figure 3: Space-time behaviour of local film thickness
(Re=40, Vg=44 m/s) along with Canny segments.

In Figure 3 result of application of Canny transform to the
data of Figure 1 is shown. Height of Canny segments has no
physical sense and is set to 0.15 just for graphical purposes.
It can be seen that Canny points accumulate on the steepest
parts of front and rear slopes of waves.
The two main parameters can be varied when applying
Canny transform to an image: amplitude threshold and the
width of filter operator. Low values of both parameters lead
to appearance of noise-generated Canny points, and high
values of these parameters lead to losing the slopes of real
waves. Optimal values of the parameters are rather hard to
establish, especially when they differ for different flow
regimes. Thus, we decided to set the minimal values of these
parameters. Minimal value was defined as the value where
the effect of parameters changing on the number of Canny
points became negligible. In this case we had to eliminate
many noise-generated Canny points, but as many as possible
Canny points, marking the slopes of waves, remain
available.

Automatic algorithm: concatenation of Canny
segments into characteristic lines

The next problem was to concatenate the obtained Canny
segments into 'Canny lines' sequences of Canny segments
on time-space surface which corresponded to characteristic
lines of certain waves. We decided to work only with Canny
segments that mark front slopes of the waves, since Canny
lines for rear slopes were essentially less stable.
Concatenation algorithm was based on spatio-temporal
behaviour of waves and the value of velocity Vcor, which
was obtained using cross-correlation analysis. As it turned
out, double value of Vcor can be safely used as the upper
estimate for velocity of any wave.
After finding a Canny segment, the adjacent moment of
time was searched for Canny segments within certain range
of distance. The latter was limited by possible velocities of
the wave. Namely, the upper estimate was 2*Vcor, and the
lower estimate was 0 m/s. If several Canny segments were
found within the range, the one corresponding to the lowest
wave's velocity was chosen. For the subsequent steps of
concatenation, the upper estimate of velocity was corrected
by substituting the current slope of characteristic line


160-

120-

z 80-

40-

0-


250-

200-
150-
z
100-

50-
0-



Figure 4
length (b


I I I
0 20 40 60
lifetime (ms)



-A


-I.


0 20 40 60
life length (mm)


' I
80


80
80


' I
100


100
100


Distribution of waves on lifetime (a) and life


It is not evident from the shape of distribution, where one
should separate the waves to primary and secondary. On that
reason an additional investigation was made, aimed to
compare the average properties of waves of certain lifetime.
As it turned out, another parameter called 'life length'
(namely, the difference between central positions of the first
and the last Canny segments in the line) was better than
lifetime, since velocities of primary waves are higher than
that of secondary waves. Fig 4 (b) shows distribution of life
length of waves. Number of waves was counted for 1
second record of 10 cm length. Bin width was 0.5 ms
(lifetime) and 1 mm (life length). Re=40, Vg=44 m/s.


L


nrrmnuA.~n,,,. -L)


'


- - -=. = _P


).






7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010


* Re=40, Vg=44 m/s




/i 0* a)


06-


1.2-




0.8-

E


0.4-




0-


0.16-



0.12-


E
E 0.08 -



0.04-
--


0.2-


I I I 1 I
20 40 60
life length, mm


I I
80 100


* Re=40, Vg=44 m/s


S
0


0 20 40 60
life length, mm


I I
80 100


I I I I
0 20 40 60
life length, mm


1.2 -1


I I II I I
0 20 40 60
life length, mm


Figure 5: Average velocity (a) and average height (b) of
waves, belonging to certain range of life length. Re=40,
Vg=44 m/s.

Figure 5 (a) shows dependence of the average velocity of
waves from certain range of 'life length'. The same
dependence for the average height of peaks of waves is
shown in Fig. 5 (b). It can be seen that these quantities
change only slightly for waves, living at least 2.5 cm of the
observed distance of 10 cm. Similar behaviour can be
observed for length of front and rear slopes of the waves
(Fig. 6 (a) and (b), respectively).
At present stage of investigation, the same value of critical
life length was used for all flow regimes in no-entrainment
conditions. This value was set equal to 2.5 cm. All the
waves that managed to travel distance longer than this value,
were regarded as primary, and processed respectively.


Figure 6: Average length of front (a) and rear slope (b) of
waves, belonging to certain range of life length. Re=40,
Vg=44 m/s.

Automatic algorithm: secondary waves processing

After selection of primary waves, search for secondary
waves, generated by these primary waves, was performed.
Any characteristic line, not marked as primary one, and
beginning close enough (less than double length of rear
slope) to the peak of a primary wave, was considered as
secondary wave, generated by this primary wave.
All the secondary waves, marked this way, were
investigated further. In particular, height of a secondary
wave and the relative coordinate of point of origin of a
secondary wave were measured.
For the latter, in some cases, additional work was required
to track the secondary wave back in time before the
beginning of characteristic line, marked by sequence of
Canny segments.
Secondary wave was considered to exist at previous
moment of time if large enough area of negative second
derivative of film thickness existed within the range of
distances of possible appearance of the wave. Additional
condition was applied that area of negative second
derivative, corresponding to secondary wave shouldn't
overlap with such area for parent primary wave. In the
earliest moment of time where the secondary wave was
considered to exist, distance between beginnings of


* Re=40, Vg=44 m/s


* a


ea@0


E
E 0.4-
j]


4-i


I I
80 100


.*0 *


* Re=40, Vg=44 m/s


I I
80 100









secondary and primary waves was measured. Relative
coordinate of inception of secondary wave was obtained by
normalizing this distance to the length of primary wave.

Results: Primary waves

Figure 7 shows dependence of primary waves velocity on
gas flow velocity. It can be seen that at high enough gas
velocities velocity of primary waves linearly grows with gas
velocity, and slightly increases with increase in liquid
Reynolds number. For Re=40 results of hand-made
processing (hmp) from Alekseenko et al. (2009b) are also
given. It can be seen that difference between hand-made
(red crosses) and automatic (solid blue circles) processing
(ap) appears only at low gas velocities.


1.6-




1.2-
-


E 0.8 -
-



0.4-




0-


Re=20
Re=40, ap
Re=40, hmp


7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010

at high gas velocities. Data on substrate thickness are also
given.
The lengths of front and back slopes of waves were defined
as follows (see Fig. 9). Peak of a wave was defined as the
first point to the left from Canny segment, where film
thickness stopped falling and began to grow. Beginning of a
wave was defined as the first point to the right from Canny
segment, where film thickness reached level of 1.1*Hsub.
Substrate thickness was defined as the most probable value
of film thickness over all the time-space record of film
thickness at fixed flow parameters. End of a wave was
defined as the first point to the left from the peak, where the
film thickness reached level of 1.1*Hsub. For any wave all
the quantities (velocity, height, length of slopes) were
measured at each instant of time, and then the average value
was fixed.

Local film thickness
+ Center of Canny segment
-- Hsub*1.1
Characteristic points of wave

0.25--
1 **.


0.2-


I I 1
20 30 40
Vgas, m/s


50
50


0.15



0.1



0.05


60
60


Figure 7: Average velocity of primary waves. Re=20, 40.


0.4-



0.3-



E
E 0.2 -



0.1


Re=40
Re=20
Re=40, Hsub
Re=20, Hsub


*
.* -(
* C


--- ** *,
****. .* *******



Lb Lf


,_\ _


x, mm


Figure 9: Definition
Vg=27 m/s.


gO


0 *

0


I I
28 30


of characteristics of waves. Re=40,


In Figure 10 dependence of average lengths of front and rear
slopes of primary waves are given. It can be seen that both
values reduce with gas velocity growth and remain nearly
* constant with change in Reynolds number. Thus, the total
Length of average primary wave is equal to 9 mm for Vg=14
0 m/s and less than 2.5 mm for Vg=52 m/s.


I I I I I
10 20 30 40
Vgas, m/s


SI
50


Figure 8: Average height of primary waves and substrate
thickness. Re=20, 40.

In Figure 8 dependence of height of primary waves on gas
velocity is shown. This value decreases with gas velocity
growth, and slightly increases with liquid Reynolds number


-


* I






7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010

correspond to Canny segments, marking secondary waves.


E
E 0.2-
I


S *


0 20 30 40
Vgas, m/s
O Re=20
S Re=40


I
50 60


b)


(11 ., .. ,. ,


E
E 0.2 -
:I


50 60
50 60


Figure 10: Average length of front (a) and rear (b) slopes of
primary waves. Re=20, 40.

Results: Secondary waves

Data on height of secondary waves are given in Figure 11
along with data on height of primary waves and the
substrate thickness. It can be seen that the average
amplitude of secondary waves is essentially lower than that
of primary waves. Besides, average amplitude of secondary
waves (i.e., difference between height of secondary waves
and substrate thickness) changes only slightly within the
studied range of gas velocities, in contrast to that of primary
waves.
Figures 12 and 13 show an example of behavior of primary
and secondary waves on x-t surface and in H vs. x plots. In
Figure 12 red corresponds to the characteristic lines of
primary waves; light green to characteristic lines of
secondary waves; orange to the areas of negative second
derivative, obtained while tracking back in time (as it was
described in Automatic algorithm: secondary waves
processing section). Film thickness is given by shades of
blue (the lighter the blue colour the higher the film
thickness). Red rectangle shows the area plotted in Figure
13.
In Figure 13 evolution of film thickness profile in six
consecutive moments of time is shown. Step functions with
height 0.1 mm correspond to Canny segments, marking
primary waves. Step functions with height 0.04 mm


Re=20, secondary waves
Re=20, primary waves
Re=20, substrate


+

S+

a


D D [1


I I I I
10 20 30 40
Vgas, mis

+ Re=4
1 Re=4
0 Re=4


I I I I I I
10 20 30 40
Vgas, m/s


I I
50 60


3, secondary waves
3, primary waves
0, substrate


Figure 11: Height of primary and secondary waves and the
substrate thickness for Re=20 (a) and Re=40 (b).

t, ms



If IIII I







8 t I I I I








Figure 12: Tracking secondary waves back in time before
the beginning of its Canny line. Re=40, Vgas44 m/s.

Step functions with height 0.06 mm correspond to the areas


O Re=20
* Re=40


S .. ........


8-



6-


E
4-



2-


-t) . . . .


0
0
o


I I "
20 30 40
Vgas, m/s


a *


5I I
50 60


W Ii









of negative second derivative, obtained by tracking the wave
back in time.


7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010

Dimensionless form of xc (Figure 14 b) was obtained by
dividing xc to total length of primary wave (i.e., distance
from the beginning to the end of primary wave).The same
results, obtained by hand-made processing, are also given
for Re=40. In accordance with Figure 13, results of
automatic processing are slightly higher than hand-made
results, and the behaviour of xc with gas velocity is the
same.


604 605 606 607 608
x, mm


607 608
x, mm


609 610


1 b)


0.8-


00.6-
n


Figure 13: Sequence of instantaneous film thickness
profiles along with markings of Canny segments for primary
(height 0.1 mm) and secondary waves (height 0.04 mm) and
areas of negative second derivative obtained while tracking
back in time (height 0.06 mm). Re=40, Vgas=44 m/s.

It can be seen that the actual moment of nucleation of
secondary wave is t=0, x=604.6 mm. This value would be
marked by hand-made processing, yielding xc (distance
from the beginning of the primary wave to the point of
origin of secondary wave) equal to 1.5 mm. The first point
marked by Canny transform for this secondary wave is t=2.5
ms, x=607.1 mm. In this case, xc would be equal to 2.3 mm.
Due to tracking the wave back in time basing on criterion of
negative H", we can automatically obtain the nucleation
point at t=l ms, x=605.7 mm (right limit of the area of
negative H"). In this case, xc=1.9 mm.
Figure 14 shows dependence of xc in dimensional (a) and
dimensionless (b) forms on Vgas. In Figure 14 (a) average
values of total length of primary waves and length of front
slope of primary waves are given. Standard deviations for
xc do not differ essentially at different Re, so the data on
standard deviation are given for Re=40 only. It can be seen
that average xc is essentially lower than total length of
primary wave, and, thus, all the secondary waves appear on
the back slopes of primary ones.


) 0 Re=20, xc
a i Re=40, xc
a Re=20, If+lb
a a Re=40, If+lb
Re=20, If
a o Re=40, If








0


I I
20 30
Vgas, m/s


40 50
40 50


Re=20
Re=40
Re=40, hmp


I I 1 1
10 20 30
Vgas, mis


T I 1
40 50


Figure 14: a) Distance from the beginning of primary wave
to the point of origin of secondary wave (xc), length of front
slope of primary wave (If) and total length of primary wave
(lf+lb). b) xc, normalized by (lf+lb), along with the results
of hand-made processing for Re=40.

On entrainment regimes: future work

Figure 15 shows an example of space-time evolution of
liquid film surface in regimes with entrainment with Canny
segments marked by red points. The wavy system in such
conditions is even more complicated than in no-entrainment
flow regimes. Due to this complexity, Canny lines are being
broken frequently for both primary and secondary waves. In
this case, some additional considerations are required to
make essential improvements in automatic algorithm.


d
O

























x. cm


Figure 15: Film thickness evolution in space and time along
with Canny segments, marking the waves. Re=350, Vg=27
m/s.

Conclusions

Regimes of annular gas-liquid flow without entrainment are
investigated with high-speed laser-induced fluorescence
technique. Automatic algorithm of data processing was
developed based on application of Canny transform. This
allowed extraction of important characteristics of primary
and secondary waves. The algorithm needs improvements
for application to entrainment regimes.

Acknowledgements

The work was supported by Federal Target Program
"Scientific and educational cadres of innovative Russia" for
2009-2013 and Russian Foundation for Basic Research
(Grant N 10-08-01145).

References

Woodmansee D.E., Hanratty T.J. Mechanism for the
removal of droplets from a liquid surface by a parallel air
flow. // Chem. Engng. Sci., 1969, V 24, p. 299-307.

Hall Taylor N.S., Nedderman R.M. The coalescence of
disturbance waves in annular two-phase flow. // Chemical
Engineering Science, 23 (1968) 551-564.

Chu K.J., Dukler A.E. Statistical characteristics of thin,
wavy liquid film. III. Structure of large waves and their
resistance to gas flow. AIChE Journal, 1975, v. 21, No 3, p.
583-593

Azzopardi B.J. Disturbance wave frequencies, velocities and
spacing in vertical annular two-phase flow. Nucl. Engng
Des., 1986, v. 92, p. 121-133.

Sekoguchi K., Mori K. New development of experimental
study on interfacial structure on gas-liquid two-phase flow.
// Exp. Heat Transfer Fluid Mech. Thermodyn. Ed. Ets, 2,
1177-88, 1997.

H. Han, Z. Zhu, K. Gabriel. A study on the effect of gas
flow rate on the wave characteristics in two-phase


7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010

gas-liquid annular flow // Nucl. Engng. Des. 236 (2006)
2580-2588

P Sawant, M. Ishii, T. Hazuku, T. Takamasa, M. Mori.
Properties of disturbance waves in vertical annular
two-phase flow // Nucl. Engng. Des., 238 (2008) pp.
3528-3541.

M. Damsohn, H.-M. Prasser. High-speed liquid film sensor
for two-phase flows with high spatial resolution based on
electrical conductance // Flow Measurement and
Instrumentation 20 (2009) 1-14.

Alekseenko S.V, Antipin VA., Cherdantsev A.V,
Kharlamov S.M., Markovich D.M. Investigation of waves
interaction in annular gas-liquid flow using high-speed
fluorescent visualization technique. // Microgravity Sci.
Technol., 2008. V20, No 3-4, pp. 271-275.

Alekseenko S.V, Antipin VA., Cherdantsev A.V,
Kharlamov S.M., Markovich D.M. Two-wave structure of
liquid film and waves interrelation in annular gas-liquid
flow with and without entrainment // Physics of Fluids,
2009 (a). Vol. 21, 061701-061704.

Alekseenko S.V, Cherdantsev A.V, Cherdantsev M.V,
Markovich D.M. Investigation of secondary waves
dynamics in annular gas-liquid flow // Microgravity
Science and Technology, 2009 (b). v. 21, Suppl. 1, pp.
221-226.

Chu K.J., Dukler A.E. Statistical characteristics of thin,
wavy liquid film. II. Studies of substrate and its wave
structure. // AIChE Journal, 20 (1974) 695-706.

Suzuki, K., Hagiwara, I., Sato, T. Heat transfer and flow
characteristics of two-phase two-component annular flow.
Int. J. Heat Mass Transfer, Vol. 26, 597-605 (1983).

Asali J.C., Hanratty T.J. Ripples generated on a liquid film
at high gas velocities. // Int. J. Multiphase Flow, 19 (1993)
229-243.

Alekseenko S.V, Cherdantsev A.V, Kharlamov S.M.,
Markovich D.M. Experimental Study of Liquid Film Wavy
Structure in Annular Two-Phase Flow. 6th International
Conference on Multiphase Flow, Leipzig, Germany, July 9 -
13, 2007, DVD-ROM Proceedings, PS5_2.

K. Ohba and K. Nagae, Characteristics and behavior of the
interfacial wave on the liquid film in a vertically upward
air-water two-phase annular flow // Nucl. Eng. Des., 1993, v.
141, p. 17-27.

Canny J. A computational approach to edge detection, IEEE
Transactions on Pattern Analysis and Machine Intelligence,
1986, Vol. PAMI-8, No. 6, pp. 679-698.




University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - Version 2.9.7 - mvs