You are on page 1of 19

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/327108617

Influence of a sudden expansion on slug flow characteristics in a horizontal


two-phase flow: a pressure drop fluctuations analysis

Article  in  Meccanica · October 2018


DOI: 10.1007/s11012-018-0892-5

CITATIONS READS

8 247

4 authors:

Abderraouf Arabi Yacine Salhi


Sonatrach University of Science and Technology Houari Boumediene
11 PUBLICATIONS   41 CITATIONS    31 PUBLICATIONS   84 CITATIONS   

SEE PROFILE SEE PROFILE

El-Khider Si-Ahmed Jack Legrand


University of Nantes University of Nantes
54 PUBLICATIONS   334 CITATIONS    355 PUBLICATIONS   8,197 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Virus induced gene silencing View project

THESIS View project

All content following this page was uploaded by Yacine Salhi on 04 August 2020.

The user has requested enhancement of the downloaded file.


Meccanica
https://doi.org/10.1007/s11012-018-0892-5 (0123456789().,-volV)
(0123456789().,-volV)

Influence of a sudden expansion on slug flow characteristics


in a horizontal two-phase flow: a pressure drop fluctuations
analysis
A. Arabi . Y. Salhi . E. K. Si-Ahmed . J. Legrand

Received: 2 December 2017 / Accepted: 14 August 2018


 Springer Nature B.V. 2018

Abstract Experimental investigations were con- 1 Introduction


ducted to study the evolution of air/water slug flow
characteristics through a horizontal sudden expansion Gas–liquid two-phase flows are encountered in a wide
having a ratio of rA = 0.444. A series of acquisition of range of different chemical reactors, power plants and
differential pressure upstream and downstream were transportation pipelines in oil and gas industry. These
carried out leading to statistical as well as spectral flows are mostly subjected to singularities, generating
analyses. The influences of both liquid and gas important head losses and affect the flow patterns.
superficial velocities on flow behavior as well as Among these singularities, sudden enlargement cor-
standard deviation were studied. Substantial modifi- responds to a common case. Most studies focused on
cations of two-phase flow distribution were reported the evolution of static wall pressure as well as the void
downstream the singularity with a quantitative reduc- fraction, both in macro-pipes (Mendler [1], Janssen
tion in standard deviation. Besides the fact that the and Kervinen [2], Velasco [3], Ahmed [4], Salhi [5])
slug flow regime didn’t persist downstream the and mini-pipes (Abdelall et al. [6], Chen et al. [7]).
singularity for low values of superficial phasic-veloc- Various theoretical and semi-empirical models, for the
ities, the latter (J‘ and Jg ) seem to have no noticeable prediction of the pressure drop, were developed. Some
effects downstream the expansion. Reduction of slug were based on momentum conservation equation
frequency between upstream and downstream the (Lottes [8], Chisholm and Sutherland [9], Ahmed
singularity was also observed. et al. [10]), the energy conservation equation
(Richardson [11], Delhaye [12]) or from the evolution
Keywords Two-phase flow  Differential pressure  of kinetic energy into static pressure (Wadle [13]).
Sudden expansion  Slug flow  PDF  PSD Some studies focused on a particular flow regime;
for instance, Aloui et al. [14, 15] investigated the
bubbly flow in the presence of a flat horizontal sudden
enlargement with an aspect ratio r = 0.444. The
pressure loss and the local parameters such as the void
A. Arabi  Y. Salhi  E. K. Si-Ahmed (&)
USTHB Physics’ Faculty Dept. Fluid Mechanics and fraction, the size and the velocity of the bubbles were
Energetics, LMFTA, Algiers, Algeria reported. They found that the evolution of the pressure
e-mail: el-khider.si-ahmed@univ-nantes.fr along a flat sudden expansion is similar to that
obtained in axisymmetrical sudden expansion. The
E. K. Si-Ahmed  J. Legrand
University of Nantes, GEPEA, CNRS, UMR-6144, 37 Bd slip velocities were, also, very high compared to the
de l’université BP406, 44602 Saint-Nazaire, France values obtained with bubbly flow in regular pipe; the

123
Meccanica

bubbles’ size decreased along their trajectories. Sch- A substantial amount of research work focusing on
midt and Friedel [16] studied a pressure loss in the slug flow has been achieved, whether in a straight pipe
presence of an annular-mist flow and developed a (Belgacem et al. [32], Al-Kayiem et al. [33]) or with
model for the prediction of singular pressure drop. the presence of singularities, such as bend (Zhao et al.
Yang et al. [17] were concerned by the flow regime [34]), orifice (da Costa Maidana et al. [35]) and
variations, in case of a vertical pipe, through a sudden T-junction (Saieed et al. [36]). However, studies on the
enlargement of r = 0.36. The authors found a transi- slug flow through a sudden enlargement are still
tion from annular to churn flow downstream the lacking. This work was undertaken to provide flow
expansion. characteristics, including statistical information, of
Of all the flow patterns encountered in field two-phase flow, especially slug flow, crossing a
operations, slug flow is dominant in horizontal and sudden expansion. It aims on examining the influence
near horizontal pipelines with and without singulari- of such singularity on this flow regime by means of its
ties (Mandhane et al. [18], Taitel and Dukler [19], Kim characteristic fluctuations. A series of differential
and Ghajar [20]). Slug flow is characterized by an pressure measurements, in a horizontal pipe, were
alternate flow of liquid and gas pockets. This inter- carried out upstream and downstream a singularity of
mittence generates a hydrodynamic instability, result- aspect ratio rA = 0.444. Statistics as well as frequency
ing in temporal fluctuations of parameters including analyses were conducted on the time series of the
local pressure, void fraction and velocity fields. differential pressure upstream and downstream the
Consequently, pipe erosion–corrosion phenomena singularity. The influence of the phasic superficial
are amplified (Thaker and Banerjee [21]); the pipe velocities was investigated.
aging is then accelerated and could possibly lead to its
destruction (Ahmed et al. [22]).
Understanding time-dependent parameter in inter- 2 Experimental facility and measurement
mittent flow, such as pressure and void fraction, is very techniques
important. Neglecting the fluctuation of these param-
eters is acknowledged to be a departure from the The experiments were performed in a two-phase
physical reality (Wang and Shodji [23]). The analysis water–air test loop sketched in Fig. 1. The same
of these fluctuations is carried out by means of experimental device was used by Belgacem et al. [32].
statistical methods as well as frequency analysis A 14 m long test facility, built in PMMA to allow
(Drahos et al. [24, 25]). The methods of deterministic optical access, was used. It is adapted to generate a
chaos analysis of wall pressure fluctuations were used gas–liquid two-phase flow concurrently. The working
by the same author [25] to characterize the intermittent fluids were water and air. The loop operates in closed
flow patterns in a horizontal pipe. The occurrence of circuit for the liquid component which is provided by a
chaotic behavior revealed well the plug–slug transi- centrifugal Noryle pump, while air is supplied by a
tion predicted by other authors. compressor. Gas flow measurements were performed
Among the parameters explored in slug flows the by two rotameters AALBORG VMRP010092 and
slug frequency, defined as the number of liquid slugs VMRP010083 type. The liquid rate was measured by
passing per unit time at a fixed position of the pipe, is an ultrasonic flow meter type GE PT878.
one of the most studied intrinsic parameter. The slug The pipe was made of several 2 m long flanged
frequency influences directly the pressure drop tubes. Care was taken to ensure that the bore was
(Greskovitch and Shrier [26]). Moreover, the intrinsic continuous (i.e. without steps at the flange positions).
frequency of this type of flow (few Hertz) can resonate Both fluids air and water entered the pipe through a
with the frequency of the pipe which can lead to the cylindrical mixing chamber. At the inlet, stratified
destruction of the latter (Kabiri-Samani et al. [27]). It flow conditions were generated by a convergent
should be noted that many correlations, for the nozzle. The test section was characterized by a sudden
prediction of the slug frequency, are reported in the area expansion with inner diameters of 0.04 m and
literature (Gregory and Scott [28], Taitel and Duckler 0.06 m respectively upstream and downstream the
[29], Zabaras [30], Fossa et al. [31]). singularity, giving an area ratio of 0.444. The singu-
larity was located at 187.5 D from the mixer.

123
Meccanica

Fig. 1 Experimental set-up

AR : Air Rotameters
C : Compressor
M : Mixer
ST: Separator Tank
WF: Water
Flowmeter
WP: Water Pump
WR: Water reservoir
ΔP : Differential
Pressure Transducer

Four-millimeter diameter pressure tapings were to the experiment’s conditions as well as the inner pipe
drilled and mounted down the pipe surface in the test diameter. Downstream the singularity (Fig. 3), for low
section. Two Freescale MPX-2010 DP differential liquid superficial velocity value, the slug flow didn’t
pressure transducers (with a range of 0–10 kPa) were persist after crossing the sudden expansion, the liquid
used. The data recordings were achieved via a Picolog slug breaks up forming a wave which does not touch
ADC20 card. the upper part of the cross section. For a majority of
The pressure drops were measured between loca- studied cases, the slug flow persists.
tions of 33 cm and 101 cm upstream and between
50 cm and 150 cm downstream the singularity. The 3.2 Pressure drop signals
sampling time was 30 s for a frequency of 200 Hz.
The exclusive use of a visual observation to determine
flow patterns has the drawback of being subjective.
3 Results and discussion Visual observation is undoubtedly a major reason of
discrepancies for researchers who have recorded
The investigations were performed for forty-two different flow regimes under essentially similar con-
 
couples j‘ ; jg of liquid and gas superficial velocities ditions, especially for high flows of gas where the
as follows: 0:239 ms  jl  0:398 ms and phases become hardly perceptible due to the strong
m m
1:061 s  jg  2:387 s . aeration of the slugs (Roman et al. [37]). Therefore,
the development of simple quantitative tools to
3.1 Flow regime map distinguish between flows was considered necessary.
In this work, visual observations were supple-
The flow pattern was first identified for each setting mented by an analysis of pressure drop fluctuations
  signals according to the method of Weisman et al.
j‘ ; jg on flow map of Mandhane et al. [18] upstream
[38]. Slug flows have typical pressure traces. Slugs can
(Fig. 2) and downstream (Fig. 3) the singularity.
be identified from the pressure signal by a pressure
Figure 2 shows that most of the experimental points
drop trace with regularly spaced peaks.
belong to the slug flow zone in the flow map of
As pointed out by Kadri et al. [39], a noteworthy
Mandhane et al. [18]. This flow pattern is distin-
phenomenon that has been observed with the present
guished from the plug flow by the presence of gas
experimental study, upstream the sudden expansion, is
bubbles in the liquid slug. This figure, despites some
the appearance of two types of slugs: regular long
slight discrepancies between the experimentally
slugs and short slugs, called hydrodynamic slug.
observed transition lines and those of Mandhane
Regular long slug can be identified from the pressure
et al. [18], the ranges of superficial gas and liquid
signal by a pressure drop trace, with regularly spaced
velocities in which this flow pattern was predominant
peaks.
were similar. These slight offsets between the
According to Weisman et al. [38], the passive
recorded points the Mandhane’s map could be related
regions between the peaks were at least twice the

123
Meccanica

Fig. 2 Flow regime map


upstream sudden expansion

Fig. 3 Flow regime map


downstream sudden
expansion

length (covered twice the time period) of the peaks. At constant upstream gas superficial velocity (Jg-
first, as we can see the number of peaks is substantially = 1.061 m/s) for three values of the liquid superficial
lower downstream compared to those recorded velocity J‘. For instance, the upstream signal in Fig. 4a
upstream the singularity for Figs. 4a–c and 5a–c. (J‘ = 0.239 m/s) is the distinctive pressure trace of
Moreover, Fig. 4a–c show signals of pressure drop slug flow as reported by Weisman et al. [38].
upstream and downstream the sudden expansion for a Nevertheless, downstream the singularity, the signal

123
Meccanica

Fig. 4 Differential pressure 5000


signals upstream and upstream
4000
downstream of the sudden downstream
expansion for different 3000 Jl = 0.239 m/s
upstream liquid superficial

Pressure drop (Pa)


velocities (Jg = 1.061 m/s) 2000

1000

-1000

-2000

-3000

-4000
0 5 10 15 20 25 30
Time (s)
(a)

5000

4000
upstream
downstream
3000 Jl = 0.292 m/s
Pressure drop (Pa)

2000

1000

-1000

-2000

-3000

-4000
0 5 10 15 20 25 30
Time (s)
(b)

5000

4000
upstream
downstream
3000 Jl = 0.398 m/s
Pressure drop (Pa)

2000

1000

-1000

-2000

-3000

-4000
0 5 10 15 20 25 30
Time (s)
(c)

123
Meccanica

Fig. 5 Differential pressure 5000


signals upstream and upstream
4000
downstream of the sudden downstream
expansion for different 3000 Jg = 1.326 m/s
upstream gas superficial

Pressure drop (Pa)


velocities (J‘ = 0.318 m/s) 2000

1000

-1000

-2000

-3000

-4000
0 5 10 15 20 25 30
Time (s)
(a)

5000

4000 upstream
downstream
3000 Jg = 1.591 m/s
Pressure drop (Pa)

2000

1000

-1000

-2000

-3000

-4000
0 5 10 15 20 25 30
Time (s)
(b)

5000

4000 upstream
downstream
3000 Jg = 2.122 m/s
Pressure drop (Pa)

2000

1000

-1000

-2000

-3000

-4000
0 5 10 15 20 25 30
Time (s)
(c)

123
Meccanica

contained small peaks with short amplitudes repre- waves do not touch the top of the pipe (Lin and
senting the passage of small waves which do not touch Hanratty [42]). According to these authors, slugs can
the wall. The non-existence of peaks revealed the non- be recognized from pseudo-slugs by examining the
persistence of the slug regime downstream the singu- amplitude of the pressure fluctuation. However, in
larity. The slugs broke down when passing the some cases, especially when gas flow rate is high, the
singularity forming small waves which do not reach distinction is not straightforward. In these cases,
the upper part of the pipe. As the superficial velocity of pseudo-slugs bridge the pipe temporarily followed
the liquid increases, the process for creating slugs by waves that do not touch the top of the pipe.
becomes stochastic (Woods et al. [40]). As a result, the Therefore, significant pressure fluctuations are
spacing between two successive slugs becomes irreg- observed, in the film liquid region. Since the blockage
ular leading to occurrence of hydrodynamic slug as is only transitory, the pressure is not sustained for the
depicted by the signals in Fig. 4b and c. An inspection whole duration of the existence of the pseudo-slug,
of Fig. 4b, c suggests a great sensitivity of flow pattern such as for slug cases.
to J‘. For values of J‘ = 0.292 m/s and J‘ = 0.398 m/s,
the peaks were more prominent downstream the 3.3 PDF signals of pressure drop
singularity revealing the persistence of the slug flow.
One should note that for J‘ equals to 0.292 m/s, a The Probability Density Functions (PDF) expresses
hydraulic jump was observed downstream the expan- the frequency occurrence of a pressure value in a time
sion. When the liquid at high velocity (Fig. 4b) flows series. Jones and Zuber [44] suggested that the flow
into a zone of lower velocity, a slight sudden rise pattern could be identified depending on the shape of
occurred at the liquid interface due to the abrupt the PDF. Costigan and Whalley [45] carried out an
deceleration of the liquid. The liquid holdup near the extensive work to link specific PDF signatures with
mixing zone of the singularity increased slowly with flow pattern configurations. The shape of PDF can be
time inducing an acceleration of the gas-phase and used as a tool to identify the flow patterns.
small disturbances at the interface of the stratified flow Figure 6a–c depict the PDF signals of a differential
were generated. While the gas holdup decreased, the pressure drop upstream and downstream the sudden
small waves increased in amplitude until a large wave expansion for three values of the liquid superficial
reached the upper wall pipe, producing a single slug of velocities while maintaining a constant gas superficial
liquid that propagated along the test section, as velocity (Jg = 1.061 m/s). For J‘ = 0.239 m/s,
revealed by the presence of peaks in the pressure drop (Fig. 6a) the shape of the PDF is characterized by a
signal. unimodal distribution with a narrow single peak
Figure 5 shows the pressure drop signals of the flow upstream the singularity. This single peak is a
upstream and downstream the expansion for different signature of a slug flow (Luo et al. [46]). Downstream
gas superficial velocities with a constant liquid the expansion, the absence of peaks revealed a
superficial velocity (J‘ = 0.318 m/s). As revealed by stratified flow regime: the slugs vanished. As the
Fig. 4a–c the number of peaks in the signal upstream liquid superficial velocity increases the pressure time
the expansion is higher than the number of peaks series exhibits a number of peaks occurring periodi-
downstream. Increasing gas superficial velocities does cally. The PDF is then characterized by a slightly
not play a major role on the slug formation compared shorter peak upstream Fig. 6b (Jg = 1.061 m/s and
to the increase of the liquid superficial velocities as J‘ = 0.292 m/s) and vice versa downstream; the peaks
reported by Vaze and Banerjee [41]. Moreover, become wider as J‘ increases Fig. 6c (Jg = 1.061 m/s
increasing Jg showed that the slug regime recorded and J‘ = 0.398 m/s). This remark does not hold when
upstream the singularity evolved into a pseudo-slug increasing Jg (see Fig. 7).
regime downstream the latter (Lin and Hanratty [42], Figure 7a–c show the PDF signals of differential
Thaker and Banerjee [43]). Upstream the singularity, pressure drop upstream and downstream the sudden
small peaks appeared on the liquid film, indicating a expansion for different gas superficial velocities for a
pseudo slug: a sub-regime of the slug flow character- constant liquid superficial velocity (J‘ = 0.318 m/s).
ized also by the intermittency of both phases with One should note, that the shape of the PDF is
liquid plugs blocking momentarily the conduct and the characterized by a unimodal distribution both

123
Meccanica

40 40

upstream upstream
35 35
downstream downstream
Jl = 0.239 m/s Jl = 0.292 m/s
30 30

25 25

PDF (%)
PDF (%)

20 20

15 15

10 10

5 5

0 0

-4000 -3000 -2000 -1000 0 1000 2000 3000 4000 5000


-4000 -3000 -2000 -1000 0 1000 2000 3000 4000 5000

Pressure drop (Pa) Pressure drop (Pa)


(a) (b)

40

upstream
35
downstream
Jl = 0.398 m/s
30

25
PDF (%)

20

15

10

-4000 -3000 -2000 -1000 0 1000 2000 3000 4000 5000


Pressure drop (Pa)
(c)
Fig. 6 PDF signals of differential pressure drop upstream and downstream of the sudden expansion for different upstream liquid
superficial velocities (Jg = 1.061 m/s)

upstream and downstream the expansion. This reveals For other couples of (J‘, Jg) the reappearance of the
the presence of a slug unit regime flow on either side of slugs downstream the enlargement is followed by an
the singularity. increase in the maximum pressure drop. Further, in
contrast with the upstream region of the expansion, the
3.4 Maximum differential pressure superficial velocity of the gas does not seem to have a
great influence on the pressure drop downstream the
The maximum pressure differences upstream and singularity.
downstream the singularity were plotted as a function Figure 9 displays the variation of the maximum
of the gas superficial velocities for different values of pressure drop as function of the liquid superficial
J‘ (Fig. 8). It is noticed that upstream the singularity, velocities for different gas superficial velocities.
the maximum pressure drop increases with the gas Upstream the singularity, an increase in the liquid
superficial velocity as reported by Luo et al. [46] in superficial velocity does not seem to have a great
case of a regular pipe. For Jg = 2.122 m/s, the effect on the evolution of the pressure drop. This is
maximum pressure drop decreases upstream the consistent with the results found by Luo et al. [46] in a
singularity before rising, this is due to the transition straight pipe for low gas phase flow velocities.
from slug flow to pseudo-slug flow. Downstream the Downstream the expansion, the increase of the
singularity, for cases where the slug flow does not superficial velocity of the liquid phase doesn’t have
persist (J‘ = 0.239 m/s and low values of Jg), the any effect on the maximum pressure drop.
maximum pressure drop is low (less than 1000 Pa).

123
Meccanica

40 40

upstream upstream
35 35
downstream downstream
Jg = 1.326 m/s Jg = 1.591 m/s
30 30

25 25

PDF (%)
PDF (%)

20 20

15 15

10 10

5
5

0
0
-4000 -3000 -2000 -1000 0 1000 2000 3000 4000 5000
-4000 -3000 -2000 -1000 0 1000 2000 3000 4000 5000
Pressure drop (Pa) Pressure drop (Pa)

(a) (b)

40

upstream
35
downstream
Jg = 2.122 m/s
30

25
PDF (%)

20

15

10

-4000 -3000 -2000 -1000 0 1000 2000 3000 4000 5000


Pressure drop (Pa)
(c)
Fig. 7 PDF signals of differential pressure drop upstream and downstream of the sudden expansion for different upstream gas
superficial velocities (J‘ = 0.318 m/s)

The maximum pressure drop results were also y is the time-averaged value and N is the number of
represented as a function of the upstream mixture instantaneous data of each run.
velocity (Jm = Jg? J‘) in Fig. 10. We see that for Figures 11 and 12 show standard deviation of the
Jm [ 2.2 m/s, the decrease in the maximum pressure pressure drop signal upstream and downstream the
drop may be regarded as an effect of the transition expansion as function of the gas and liquid superficial
from slug flow to pseudo-slug flow or another sub- velocities respectively. It appears from Fig. 11 that the
regime of slug flow (Thaker and Banerjee [21]). increase in Jg is accompanied by an increase in the
standard deviation, behavior already observed in
3.5 Standard deviation of differential pressure straight pipes (Luo et al. [46]). For a gas superficial
velocity of Jg = 2.122 m/s, the standard deviation is
The standard deviation is a statistical parameter used reduced due to the flow transition. Downstream the
to evaluate the intensity of the dispersion. The singularity, for J‘ = 0.239 m/s, the increase in gas flow
leads to an increase in the standard deviation. For the
standard deviation is computed as STD ¼
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi other cases, studied, the same remarks made for the
PN
1
N i¼1 ðyi  y Þ2 where yi is the instantaneous value, maximum pressure parameter are to be drawn i.e. the

123
Meccanica

Fig. 8 Evolution of the 9000


maximum differential
pressure as a function of the Jl = 0.239 m/s Jl = 0.292 m/s Jl = 0.398 m/s
8000
upstream gas superficial upstream upstream upstream
velocities for different downstream downstream downstream
upstream liquid superficial 7000

Maximum pressure drop (Pa)


velocity
6000

5000

4000

3000

2000

1000

0
1,0 1,2 1,4 1,6 1,8 2,0 2,2 2,4
Jg (m/s)

Fig. 9 Evolution of the 9000


maximum differential
pressure as a function of the Jg = 1.326 m/s Jg = 1.591 m/s Jg = 2.122 m/s
8000
upstream liquid superficial upstream upstream upstream
velocities for different downstream downstream downstream
upstream gas superficial 7000
Maximum pressure drop (Pa)

velocities
6000

5000

4000

3000

2000

1000

0
0,25 0,30 0,35 0,40
Jl (m/s)

superficial velocity does not seem to have a great downstream the singularity. This is due to the complex
influence on the dispersion of the pressure drop signal. nature of the waves, resulting from the breaking of the
An increase in the liquid flow rate (Fig. 12) causes liquid slug at the passage of the sudden enlargement. A
an increase in the dispersion of the pressure drop thorough investigation in terms of waves would be
signal upstream the expansion. Nonetheless, the liquid interesting to conduct to get a better understanding on
superficial velocity influence seems uncertain the influence of both liquid and gas superficial

123
Meccanica

Fig. 10 Evolution of the 9000


maximum differential
pressure as a function of the 8000 Upstream
upstream mixing velocity Downstream
Jm = Jg? J‘

Maximum pressure drop (Pa)


7000

6000

5000

4000

3000

2000

1000

0
1,2 1,4 1,6 1,8 2,0 2,2 2,4 2,6 2,8 3,0
J m (m/s)

Fig. 11 Evolution of the 1200


standard deviation of the
differential pressure as a 1100 Jl = 0.239 m/s Jl = 0.292 m/s Jl = 0.398 m/s
function of the upstream gas upstream upstream upstream
1000
superficial velocities for downstream downstream downstream
different upstream liquid 900
superficial velocities
Standard deviation (Pa)

800

700

600

500

400

300

200

100

0
1,0 1,2 1,4 1,6 1,8 2,0 2,2 2,4
Jg (m/s)

velocities on the flow downstream the sudden extracting the frequencies present in a system. In
expansion. two-phase flow, the PSD can be used to distinguish
between the flow regimes transition (Hanafizadeh
3.6 Spectral analysis et al. [47]) and to determine the slug frequency
(Heywood and Richardson [48]). The PSD, in this
It is common to use Power Spectral Density (PSD) to study, was obtained with applied the Fast Fourier
study transient phenomena. The PSD allows Transform (FFT) algorithm on the autocorrelation

123
Meccanica

Fig. 12 Evolution of the 1200


standard deviation of the
differential pressure as a 1100 Jg = 1.326 m/s Jg = 1.591 m/s Jg = 2.122 m/s
function of the upstream gas upstream upstream upstream
1000
superficial velocities for downstream downstream downstream
different upstream liquid 900
superficial velocities

Standard deviation (Pa)


800

700

600

500

400

300

200

100

0
0,25 0,30 0,35 0,40
Jl (m/s)

function of time series of pressure drop. The superficial velocities with a fixed liquid superficial
equation of PSD is given by Pxx(f) = $T0 Rxx(s)e-j2pfsds, velocity. As Jg increases the number of peaks expands
where Pxx(f) is the power spectrum density function indicating that with the transition to the pseudo-slug,
in the frequency domain, s represents time delay, the flow becomes more chaotic.
T is the sampling time and Rxx is the auto-correlation Comparing the dominant frequencies (frequency
function for a given time delay given by corresponding to the maximum amplitude) upstream
Rxx ðsÞ ¼ limT!1 T1 r T0 xðtÞxðt þ sÞdt, x(t) is the instan- and downstream, it appears in Figs. 13 and 14 that the
taneous value of pressure drop collected at time t, frequency does not vary much (difference less than
between 0 and T. 10%). However, this frequency downstream the
Figure 13 shows the PSDs obtained, upstream and expansion represents both the slugs and the large
downstream, from the pressure drop signals for waves (which do not touch the top of the wall)
different liquid superficial velocities at Jg = 1061 m/ resulting from the breaking of the slugs while passing
s. It is noted that in all cases the frequency spectrum is through the singularity. Consequently, the PSD cannot
mainly concentrated below 10 Hz and the amplitude be used in this case to characterize the slug frequency
of the PSD is less important downstream than the one downstream the sudden enlargement.
upstream the singularity where the spectrum of
frequencies exhibits a broad band of frequencies 3.7 Slug frequency
characteristic of chaotic systems (Cai et al. [49]).
Downstream, this phenomenon is less present, which Figures 15 and 16 depict the variation in the slug
suppose by crossing the sudden expansion the flow frequency upstream and downstream the expansion for
becomes less chaotic. Except for the case of J‘- different liquid and gas superficial velocities, respec-
= 0.239 m/s and Jg = 1.061 m/s (Fig. 13a) where the tively. The slug frequencies upstream were obtained
slug flow is not present downstream the singularity. by extracting the dominant frequency of the PSD.
The spectrum in this case is characterized by a band Downstream, on the other hand, the counting method
with no peaks, indicating that the system is hazardous was used (Hubbard [51], Belgacem et al. [32]). The
(Xu et al. [50]). Figure 14 shows the PSDs obtained method consists in counting the number of peaks (N)
from the differential pressure signals for different gas of a pressure signal, dividing it by the acquisition time

123
Meccanica

50000 50000
Jl = 0.239 m/s Jl = 0.239 m/s
Upstream Downstream
40000 40000

30000 30000
PSD

PSD
20000 20000

10000 10000

0 0
0 5 10 15 0 5 10 15
Frequency (Hz) (a) Frequency (Hz)

100000 100000
Jl = 0.292 m/s Jl = 0.292 m/s
Upstream Downstream
80000 80000

60000 60000
PSD
PSD

40000 40000

20000 20000

0 0
0 5 10 15 0 5 10 15
Frequency (Hz) Frequency (Hz)
(b)
100000 100000
Jl = 0.398 m/s Jl = 0.398 m/s
Upstream 80000 Downstream
80000

60000 60000
PSD
PSD

40000 40000

20000 20000

0 0
0 5 10 15 0 5 10 15
Frequency (Hz) (c) Frequency (Hz)

Fig. 13 PSD of differential pressure drop upstream and downstream of the sudden expansion for different upstream gas superficial
velocities (Jg = 1.061 m/s)

(t) to obtain the frequency of the slugs, fs ¼ Nt . singularity. These results do not confirm the principle
Obviously, according to the above analysis, a decrease of the persistence of the frequency of the slugs through
of the number of slugs downstream was observed the singularities stated by Azzopardi et al. [52]. The
compared to those recorded upstream the singularity. rupture of the slugs is explained by the fall of the liquid
Hence, the sudden expansion played the role of slug holdup due to the expansion of the diameter of the pipe
catcher allowing only the fast ones to cross the (from 40 mm to 60 mm).

123
Meccanica

100000 100000
Jl = 1.326 m/s
J g = 1.326 m/s Downstream
80000 Upstream 80000

60000 60000

PSD
PSD

40000 40000

20000
20000

0
0 0 5 10 15
0 5 10 15 (a)
Frequency (Hz)
Frequency (Hz)
150000
150000 Jl = 1.591 m/s
Jg = 1.591 m/s
Downstream
Upstream

100000
100000

PSD
PSD

50000
50000

0
0 0 5 10 15
0 5 10 15
(b) Frequency (Hz)
Frequency (Hz)
100000
100000
Jg = 2.122 m/s Jg = 2.122 m/s
Upstream 80000 Downstream
80000

60000 60000
PSD
PSD

40000 40000

20000 20000

0 0
0 5 10 15 0 5 10 15
Frequency (Hz) (c) Frequency (Hz)

Fig. 14 PSD of differential pressure drop upstream and downstream of the sudden expansion for different upstream gas superficial
velocities (Jl = 0.318 m/s)

Upstream the abrupt expansion, it was found that superficial velocity showed any influence on the slug
the gas superficial velocity has no influence on the frequency.
frequency of the slugs, whereas the increase in the Figure 17 depicts the results of the liquid slugs
liquid superficial velocity leads to an increase in the frequencies based on the Strouhal number Stg ¼ fJs Dg (fs,
slug frequency as reported by Wang et al. [53] for a the slug frequency, D is the pipe internal diameter) as a
regular horizontal channel. Downstream the expan-
sion, neither the liquid superficial velocity nor the gas function of the Lockhart–Martinelli’s parameter X ¼

123
Meccanica

3,5

Jl = 0.239 m/s Jl = 0.292 m/s Jl = 0.398 m/s


3,0 upstream upstream upstream
downstream downstream downstream
2,5
Slug frequency (Hz)

2,0

1,5

1,0

0,5

0,0
1,0 1,2 1,4 1,6 1,8 2,0 2,2 2,4
Jg (m/s)

Fig. 15 Evolution of the slug frequency as a function of the upstream gas superficial velocities for different upstream liquid superficial
velocities

Fig. 16 Evolution of the 3,5


slug frequency as a function
of the upstream liquid Jg = 1.326 m/s Jg = 1.591 m/s Jg = 2.122 m/s
superficial velocities for 3,0 upstream upstream upstream
different upstream gas
downstream downstream downstream
superficial velocities
2,5
Slug frequency (Hz)

2,0

1,5

1,0

0,5

0,0
0,25 0,30 0,35 0,40
Jl (m/s)

123
Meccanica

Fig. 17 Gas based Strouhal 1


number against Lockhart–
Martinelli Upstream
Downstream

Gas based Strouhal number Stg


0,1

0,01

1E-3
4 6 8 10 12
Lockhart-Martinelli parameter X

rffiffiffiffiffiffiffiffi qffiffiffiffi
ðdP
dx Þ‘
  drop and standard deviation upstream the sudden
¼ qq‘ JJg‘ ( dP
dx i and qi are respectively the expansion. While, the liquid superficial velocity
ð dx Þg
dP
g

phasic pressure gradient and density). Indeed, Kaji seems to have no effect, at least for the range of
et al. [54] showed a linear correlation for the case of velocities considered within this study. Nonethe-
straight pipe. Obviously a linear fit holds in the less J‘ and Jg seem to have no influence down-
upstream case while downstream a greater dispersion stream the singularity.
was recorded. • Slug frequency decreases downstream the sudden
expansion.

4 Concluding remarks Compliance with ethical standards

Conflict of interest The authors declare that they have no


The effect of an abrupt expansion on the slug regime conflict of interest.
behavior in a horizontal pipe was examined through
differential pressure fluctuations. Visual observations
were supplemented by an analysis of pressure drop References
fluctuations signals according to the method of Weis-
man et al. [38]. The influence of both fluids superficial 1. Mendler OJ (1966) Sudden expansion losses in single and
velocities was carried out upstream and downstream two-phase flow. Dissertation, University of Pittsburgh,
the singularity. The following remarks hold: Pennsylvania
2. Janssen E, Kervinen JA (1966) Two-phase pressure drop
• The non-persistence of the slug regime down- across contractions and expansions of water–steam mixture
at 600 to 1400 psia. In: Report Geap 4622-1965-US
stream the enlargement was recorded for low
3. Velasco I (1975) L’écoulement diphasique à travers un
liquid and gas superficial velocities. However, for élargissement brusque. Dissertation, Catholic University of
higher values of liquid and gas superficial veloc- Louvain, Belgium
ities the persistence of the slug regime was 4. Ahmed WE (2005) Two-phase flow through sudden area
expansions. Dissertation, McMaster University. Hamilton.
observed.
Ontario. Canada
• The gas superficial velocity has a direct effect on 5. Salhi Y (2010) Contributions théorique et expérimentale à
the flow behavior as well as the maximum pressure l’étude des phénomènes de transition d’un écoulement

123
Meccanica

stratifié à l’écoulement poche/bouchon dans une conduite 25. Drahos J, Tihon J, Serio C, Lübbert A (1996) Deterministic
horizontale en présence de singularité. Dissertation, chaos analysis of pressure fluctuations in a horizontal pipe at
USTHB Algiers Algeria intermittent flow regime. Chem Eng J 64:149–156
6. Abdelall FF, Hahn G, Ghiaasiaan SM, Abdel-Khalik SI, 26. Greskovich EJ, Shrier AL (1971) Pressure drop and holdup
Jeter SS, Yoda M, Sadowski DL (2005) Pressure drop in horizontal slug flow. AIChE J 17:1214–1219
caused by abrupt flow area changes in small channels. Exp 27. Kabiri-Samani AR, Borghei SM, Saidi MH (2007) Fluctu-
Thermal Fluid Sci 29:425–434 ation of air–water two-phase flow in horizontal and inclined
7. Chen IY, Liu CC, Chien KH, Wang CC (2007) Two-phase water pipelines. J Fluids Eng 129(1):1–14
flow characteristics across sudden expansion in small rect- 28. Gregory GA, Scott DA (1969) Correlation of liquid slug
angular channels. Exp Thermal Fluid Sci 32:696–706 velocity and frequency in horizontal concurrent gas–liquid
8. Lottes PA (1961) Expansion losses in two-phase flow. Nucl slug flow. AIChE J 42:901–907
Sci Eng 9:26–31 29. Taitel Y, Dukler AE (1977) A model for slug frequency
9. Chisholm D, Sutherland LA (1969) Prediction of pressure during gas-liquid flow in horizontal and near-horizontal gas-
gradient in pipeline systems during two-phase flow. Proc liquid flow. Int J Multiph Flow 19(5):829–838
Inst Mech Eng 184(Pt 3C):24–32 30. Zabaras G (2000) Prediction of slug frequency for gas/liquid
10. Ahmed WH, Ching CY, Shoukri M (2007) Pressure flows. SPE 65093 SPE J 5 (3)
recovery of two-phase flow across sudden expansions. Int J 31. Fossa M, Guglielmini G, Marchitto A (2003) Intermittent
Multiph Flow 33:575–594 flow parameters from void fraction analysis. Flow Meas
11. Richardson B (1958) Some problems in horizontal two- Instrum 14:161–168
phase, two-component flow. Report ANL-5949 32. Belgacem I, Salhi Y, Hammoudi M, Si-Ahmed E-K,
12. Delhaye JM (1981) Singular pressure drops in two-phase Legrand J (2015) Development and statistical characteri-
and heat transfer in the power and process industries. Edited zation of slug in two-phase flow along horizontal pipeline.
Bergles AE. Hemisphere, Washington Mech Ind 16:307
13. Wadle M (1989) A new formula for the pressure recovery in 33. Al-Kayiem HH, Mohmmed AO, Al-Hashimy ZI, Time RW
an abrupt diffuser. Int J Multiph Flow 15(2):241–256 (2017) Statistical assessment of experimental observation
14. Aloui F, Souhar M (1996) Experimental study of a two- on the slug body length and slug translational velocity in a
phase flow bubbly flow in a duct symmetric sudden horizontal pipe. Int J Heat Mass Transf 105:252–260
expansion-Part I. Visualization, pressure and void fraction. 34. Zhao D, Omar R, Abdulkadir M, Abdulkareem LA, Azzi A,
Int J Multiph Flow 22(4):651–665 Saidj F, Hernandez-Perez V, Hewakandamby BN, Azzo-
15. Aloui F, Souhar M (1996) Experimental study of a two- pardi BJ (2017) The control and maintenance of desired
phase flow bubbly flow in a duct symmetric sudden flow patterns in bends of different orientations. Flow Meas
expansion-Part II. Liquid and bubble velocities bubble size. Instrum 53:230–242
Int J Multiph Flow 22(5):849–861 35. da Costa Maidana N, Rosa ES (2018) Flow disturbances
16. Schmidt J, Friedel L (1996) Two-phase pressure change induced by an orifice plate in a horizontal air-water flow in
across sudden expansions in duct areas. Chem Eng Commun the slug regime. Exp Thermal Fluid Sci 94:59–76
141:175–190 36. Saieed A, Pao W, Hewakandamby B, Azzopardi BJ, Wood
17. Yang Y, Li G, Zhou F, Chen X (2001) The effect of sudden DA, Ali HM (2018) Experimental investigation on the
change in pipe diameter on flow patterns of air-water two- effect of diameter ratio on two-phase slug flow separation in
phase flow in vertical pipe (II) sudden expansion cross a T-Junction. J Petrol Sci Eng 170:139–150
section. Chin J Chem Eng 9(2):221–223 37. Roman AJ, Kreitzer PJ, Ervin JS, Hanchak MS, Byrd LW
18. Mandhane JM, Gregory GA, Aziz K (1974) A flow pattern (2016) Flow pattern identification of horizontal two-phase
map for gas-liquid flow in horizontal pipes. Int J Multiph refrigerant flow using neural networks. Int Commun Heat
Flow 1(4):537–553 Mass Transf 71:254–264
19. Taitel Y, Dukler AE (1976) A model for predicting flow 38. Weisman J, Duncan D, Gibson J, Crawford T (1979) Effects
regime transitions in horizontal and near-horizontal gas– of fluid properties and pipe diameter on two-phase flow
liquid flow. AIChE J 22:47–55 patterns in horizontal lines. Int J Multiph Flow
20. Kim D, Ghajar AJ (2002) Heat transfer measurements and 5(6):437–462
correlations for air–water flow of different flow patterns in a 39. Kadri U, Zoeteweij ML, Mudde RF, Oliemans RVA (2009)
horizontal pipe. Exp Thermal Fluid Sci 25:659–676 A growth model for dynamic slugs in gas–liquid horizontal
21. Thaker J, Banerjee J (2016) Influence of intermittent flow pipes. Int J Multiph Flow 35(5):439–449
sub-patterns on erosion-corrosion in horizontal pipe. 40. Woods BD, Fan Z, Hanratty TJ (2006) Frequency and
J Petrol Sci Eng 145:298–320 development of slugs in a horizontal pipe at large liquid
22. Ahmed WH (2012) Nuclear power-practical aspects. In: flows. Int J Multiph Flow 32(8):902–925
InTech, New York, USA, pp 153–178. Chapter 6 41. Vaze MJ, Banerhee J (2012) Prediction of liquid height for
23. Wang S, Shoji M (2002) Fluctuation characteristics of two- onset of slug flow. Can J Chem Eng 90(5):1295–1303
phase flow splitting at a vertical impacting T-junction. Int J 42. Lin PY, Hanratty TJ (1987) Detection of slug flow from
Multiph Flow 28:2007–2016 pressure measurements. Int J Multiph Flow 13(1):13–21
24. Drahos J, Cermak J, Seluck K, Ebner L (1987) Characteri- 43. Thaker J, Banerjee J (2015) Characterization of two-phase
zation of hydrodynamic regimes in horizontal two-phase slug flow subregimes using flow visualization. J Petrol Sci
flow part II: analysis of wall pressure fluctuations. Chem Eng 135:561–576
Eng Process 22:45–52

123
Meccanica

44. Jones OC, Zuber N (1975) The interrelation between void patterns and transitions in a small, horizontal, rectangular
fraction fluctuations and flow patterns in two-phase flow. Int channel. J Fluids Eng 118:383–390
J Multiph Flow 2(3):273–306 50. Xu J, Bao X, Wei W, Shi G, Shen S, Bi HT, Grace JR, Lim
45. Costigan G, Whally PB (1997) Slug flow regime identifi- CJ (2004) Statistical and frequency analysis of pressure
cation from dynamic void fraction measurements in vertical fluctuations in spouted beds. Powder Technol 140:141–154
air–water flows. Int J Multiph Flow 23(2):263–282 51. Hubbard MG (1965) An analysis of horizontal gas–liquid
46. Luo XM, He LM, Lu YL (2010) Fluctuations characteristics slug flow. Dissertation, University of Houston TX
of gas-liquid two-phase slug flow in horizontal pipeline. AIP 52. Azzopardi B, Ijioma A, Yang S, Lokman LA, Azzi A,
Conf Proc 1207:162–171 Abdulkadir M (2014) Persistence of frequency in gas-liquid
47. Hanafizadeh P, Eshraghi J, Alireza Taklifi A, Ghanbarzadeh flows across a change in pipe diameter or orientation. Int J
S (2016) Experimental identification of flow regimes in gas– Multiph Flow 67:22–31
liquid two-phase flow in a vertical pipe. Meccanica 53. Wang X, Guo L, Zhang X (2007) An experimental study of
51(8):1771–1782 the statistical parameters of gas-liquid two-phase slug flow
48. Heywood NI, Richardson JF (1979) Slug flow of air-water in horizontal pipeline. Int J Heat Mass Transf 50:2439–2443
mixtures in a horizontal pipe: determination of liquid 54. Kaji R, Azzopardi BJ (2009) Investigation of flow devel-
holdup by -c-ray absorption. Chem Eng Sci 34:17–30 opment of co-current gas–liquid vertical slug flow. Int J
49. Cai Y, Wambsganss MW, Jendrzejczyk JA (1996) Appli- Multiph Flow 35:335–348
cation of chaos theory in identification of two-phase flow

123

View publication stats

You might also like