IV Journeys in Multiphase Flows (JEM 2017)
March 27-31, 2017, São Paulo, SP, Brazil
Copyright © 2017 by ABCM
Paper ID: JEM-2017-0026
EXPERIMENTAL MEASUREMENTS OF HORIZONTAL THREE-PHASE
SOLID-LIQUID-GAS SLUG FLOW WITH HYDRATE-LIKE PARTICLES
Luis M.M. Rosasa, [email protected]
Carlos L. Bassania, [email protected]
Moisés A.M. Netoa, [email protected]
Marco J. da Silvaa, [email protected]
Rigoberto E.M. Moralesa, [email protected]
Amadeu K. Sumb, [email protected]
a
Multiphase Flow Research Center (NUEM), Federal University of Technology – Paraná (UTFPR), R. Deputado Heitor Alencar
Furtado, 5000, Bloco N, Campo Comprido, Curitiba/PR, CEP 81280-340.
b
Hydrates Energy Innovation Laboratory, Chemical and Biological Engineering Department, Colorado School of Mines, 1500
Illinois St., Golden, CO 80401, USA.
Abstract. Gas hydrate is a main flow assurance concern for worldwide oil companies due to the high risk of pipe
blockages. Those blockages represent a global obstacle to the successful production of deep-water hydrocarbons.
Hydrates are crystals formed by the trapping of gas molecules into cages formed by hydrogen-bonded water molecules
and may form at the water-gas interface. Right after the hydrate formation onset, when the volumetric fraction and the
size of the particles are still small, the particles flow homogeneously dispersed in the liquid phase. However, the
particles may interact with the gas-liquid flow, changing the flow hydrodynamics. For offshore operations, the phases
are assumed as flowing predominantly in the slug flow pattern due to the range of gas and liquid superficial velocities
in those operations. Understanding the effects of the solid particles introduction in the slug flow is essential to improve
the efficiency and safety of oil and gas facilities. The purpose of the present work is to experimentally characterize
solid-liquid-gas slug flow with the presence of homogeneously dispersed hydrate-like particles. Experimental tests
were carried out with polyethylene particles of 0.5 mm-diameter with density similar to the hydrates (938 kg/m3). The
test section comprised a 26 mm-ID, 9 m-long horizontal duct made of transparent Plexiglass. High Speed Imaging was
used to analyze the bubbles shape behavior due to the introduction of the solid particles. Resistivity sensors were
placed in the test section to measure the unit cell translational velocity, the slug flow frequency and the bubble and
slug region lengths. Two distinct solid particles concentrations were tested (6 and 8 g/dm3) and compared to a similar
case of liquid-gas slug flow.
Keywords: flow assurance, hydrates, three-phase solid-liquid-gas slug flow, high speed imaging, resistivity sensor.
1. INTRODUCTION
The slug flow pattern is often assumed as the prevailing one in oil and gas offshore exploitation processes due to the
characteristic operation ranges of the superficial velocities of the phases. Slug flows are characterized by the
intermittent succession of two bodies: a liquid slug, which may or may not contain dispersed gas bubbles in it, and an
elongated bubble sliding over a thin liquid film. Together, those two structures constitute that what is known as a unit
cell (Shoham, 2006). The slug and the elongated bubble possess characteristic velocities and geometric features such as
lengths and phase fractions. Those characteristics depend on time and space and their prediction is relevant in the
design of offshore facilities.
Besides oil and gas, offshore operations may also contain water (brine) and solid particles dispersed in the mixture,
such as sand. High water cut, high pressure and low temperature conditions – often found in offshore exploitation
scenarios – favor the formation of a new phase, known as hydrate. Hydrates are crystals formed by the trapping of gas
molecules into cages formed by hydrogen-bonded water molecules (Sloan and Koh, 2008) and are related to pipe
blockage risks, with consequent income losses caused by production interruptions. Hydrates may form at the gas-water
interface, where a more effective contact between the phases occurs (Sloan et al., 2011), thus forming a dispersion. The
dispersion may be homogeneous or heterogeneous, depending on the concentration and size of the particles (Peker and
Helvaci, 2007). At some point, the particles may agglomerate due to capillary and intermolecular forces (Camargo et
al., 2000), forming: (i) a moving or stationary bed, related to flow restriction; or (ii) a massive plug, which will block
the production.
Pipe blockage due to hydrate formation is one of the main challenges faced by oil and gas production companies,
especially when dealing with deep and cold waters (Cardoso et al., 2015). Efforts have been made on experimentally
characterizing and modelling gas-liquid slug flows (Bassani et al., 2016; Castillo, 2013; Kjeldby et al., 2013; Rogero,
2009; Shoham, 2006; Taitel and Barnea, 1990) and solid-liquid flows (Doron et al., 1987; Doron and Barnea, 1993;
Peker and Helvaci, 2007). However, studies on three-phase solid-liquid-gas flows are scarce and focused in the flow
L.M.M. Rosas, C.L. Bassani, M.A. Marcelino Neto, M.J. da Silva, R.E.M. Morales, A.K. Sum
Experimental Analysis of Horizontal Three-Phase Solid-Liquid-Gas Slug Flows with Hydrate-Like Particles
effects on the transportation of the solid particles (Goharzadeh et al., 2010; Goharzadeh and Rodgers, 2009; Stevenson
and Thorpe, 2003, 2002), but do not comprise the effects of the particles in the flow hydrodynamic characteristics, such
as the slug flow frequency and the unit cell geometry. Those works also assume particles with density higher than
water, such as sand. However, this is not the case of hydrates, which are often lighter than water and thus tend to
fluctuate (Sloan and Koh, 2008).
The present study aims to experimentally characterize the gas-liquid-solid three-phase flow in a horizontal pipe. The
slug flow pattern was chosen, since it is the prevailing one in offshore petroleum operations. To avoid experimentation
with real hydrate particles – which would demand a long distance flowloop, with a precise temperature and pressure
monitoring system – polyethylene particles were chosen to represent the hydrates, keeping a similar density and size
(938 kg/m3 with 0.5 mm diameter). The objective is to understand the effects introduced by the solid particles on the
main parameters of the slug flow (namely, unit cell translational velocity, slug flow frequency and unit cell region
lengths) by means of resistive sensors and High Speed Imaging in a 9 m-length transparent flowloop with 26 mm-ID.
2. EXPERIMENTAL FLOWLOOP
An experimental flowloop was assembled to characterize three-phase solid-liquid-gas horizontal slug flows using
High Speed Imaging and resistive sensors. Table 1 presents the experimental flowloop characteristics and the range of
parameters covered by the measurements. The fluids used were air and water at ambient conditions (approximately
100 kPa and 298 K). A total of 8 liquid-gas superficial velocities combinations was measured for two different solid
particles concentration (6 and 8 g/dm3). Two-phase liquid-gas flow – that is, without the solid particles – was measured
at the same conditions for comparison purposes. Each measure was made three times to assure repetition. All the jL/jG
combinations fall within the slug flow pattern, visually confirmed since the pipeline is made of transparent Plexiglass.
Table 1. Experimental flowloop characteristics and range of measured parameters.
Fluids
Air and water, at slug flow pattern
Particles material/density/diameter
Polyethylene / 938 kg/m3 / 0.5 mm
Pipe wall material
Plexiglass, transparent
Pipeline inclination
Horizontal
Flowloop length
9 m (~346D)
Flowloop pressurization
Ambient
Flowloop temperature
Ambient, isothermal
Liquid superficial velocity range
0.5 ≤ jL ≤ 1.5 m/s
Gas superficial velocity range
0.25 ≤ jG ≤ 1.5 m/s
Solid particles concentration
0 / 6 / 8 g/dm3
Mixture superficial velocity range
1 ≤ J ≤ 2 m/s
8 jL/jG combinations x 3 particles concentration
Number of measures
x 3 repetitions each = 72 measurements
Figure 1 shows a scheme of the experimental flowloop. Water is stored in a tank (i) of 250 dm3, where the solid
particles are added at the desired concentration. A mixer (ii) is placed inside the tank to assure a homogeneous
dispersion. The solid-liquid dispersion passes through a centrifugal pump (iii) for up to 200 kPa, which is fed by a 3 hp
electric motor (iv). The water flow rate is controlled by a variable-frequency drive (v) actuating on the electric motor. A
Coriolis-type flowmeter (vi) is positioned downstream of the pump and upstream of the mixing location with the gas
phase. In addition to the water flow rate, the Coriolis flowmeter returns the water density and its temperature. The water
dynamic viscosity is estimated using the temperature value via an empirical correlation (Fox et al., 2011).
In a parallel branch of the flowloop, air is compressed (vii) up to 800 kPa. The air is stored in two pressure
vessels (viii) with working pressure of 1.4 MPa and combined capacity of 600 dm3. The gas flow rate is measured
through an orifice plate (ix). A differential pressure transducer (x), designed for a measurement range of 0 to 60 kPa,
evaluates the pressure drop in the orifice plate. The pressure difference is correlated to the volumetric flow rate via
calibration with a rotameter, with a maximum deviation of 2.9%. Additionally, the gas density and dynamic viscosity
are evaluated using empirical correlations (Fox et al., 2011) as a function of the local air temperature and gauge
pressure measured at the orifice plate inlet (xi). The gas flow rate is controlled by a manual valve positioned
downstream of the orifice plate (xii).
The phases are mixed at the inlet of the horizontal pipeline via a parallel plate mixer (xii), which locally generates a
stratified flow pattern. This pattern should naturally evolve to slug flow if the phases superficial velocities are
propitious, which was indeed the case for all the jL/jG combinations tested in this work. In order to ensure this flow
pattern transition, a development section length of 5.2 m (~200D) was used before the measurement points. Resistivity
sensors (xiii) were positioned at 5.2, 6.5 and 7.8 m (~200 / 250 / 300D) of the pipe inlet. Gauge pressure
transducers (xiv) were inserted near each resistive sensor. A High Speed Camera (xv) was positioned at 6.45 m (~248D)
from the pipe inlet. A pipe length of 1.2-m (~46D) was inserted after the last resistive sensor to avoid any reverse
IV Journeys in Multiphase Flows (JEM 2017)
upstream effects. The flowloop has a total length of approximately 9 m. The fluids are discharged back to the water
tank, where the separation of the phases occurs naturally by gravity. The air is released to the ambient, while the solidliquid dispersion is reinjected on the flowloop.
9 m (~346D)
Differential pressure
transducer (x)
7.8 m (~300D)
1.2 m
(~46D)
6.5 m (~250D)
Gauge pressure
transducer (xi)
Gas valve
(xii)
Resistive sensors
(xiii)
5.2 m (~200D)
Parallel plate
mixer (xii)
Orifice plate
(ix)
Coriolis-type
flowmeter (vi)
High Speed
Camera (xv)
Gauge pressure
transducers (xiv)
6.45 m (~248D)
Variable
frequency
drive (v)
Pump (iii)
Electric motor
(iv)
Water tank (i)
Particles mixer (ii)
Figure 1. Experimental flowloop.
The High Speed Camera is used to visualize the flow pattern. The acquisition rate can be set for up to 3600 frames/s
at its higher resolution of 1024x1024 pixels. To avoid light refraction, a box of Plexiglas – full with water – was
positioned so as to wrap the pipe, as shown in Figure 2. A light source was used to illuminate the ambient. A diffuser
surface was placed between the light source and the Plexiglas box to enhance image contrast.
The resistive sensors measurements follow the method proposed by Machado et al. (2013). The resistive sensor,
shown in Figure 3a, is composed by a printed circuit board (PCB) made of glass fiber, containing two wires made of
stainless steel, with 0.12 mm of diameter and separated by 3 mm each to another. One wire acts as an electric current
transmitter, while the other acts as a receiver. The transmitter wire is fed by a squared electrical signal with amplitude
of ±5V and frequency of 1.75 kHz. The electrical resistance has a linear dependence on the amount of liquid between
the wires, thus allowing measuring the liquid height hL in the cross section as (Castillo, 2013):
hL V t   VL

D VG  VL
(1)
being V(t) the electric tension signal of the resistive sensor and VL and VG the electric tensions when the pipe cross
section is full of liquid and gas, respectively. Assuming a flat liquid-gas interface along the cross sectional area, as
shown in Figure 3b, the gas fraction RG can be related to the liquid height hL as:
RG  1 
1 
 2h
arccos  1  L
D



2
  2hL 
 2hL  
1
1
1




 



D 
D  
 


(2)
L.M.M. Rosas, C.L. Bassani, M.A. Marcelino Neto, M.J. da Silva, R.E.M. Morales, A.K. Sum
Experimental Analysis of Horizontal Three-Phase Solid-Liquid-Gas Slug Flows with Hydrate-Like Particles
Light source
Diffuser surface
Flow direction
Plexiglas
visualization box
High Speed
Camera
Figure 2. High Speed Camera positioned at the flowloop, with a Plexiglas box to avoid light refraction and a lightdiffuser surface to enhance image contrast.
Three resistive sensors are always coupled together in a set at each measurement point of the test section, as shown
in Figure 3c. The center plate acts as grounding so as to avoid interferences. The signals of the other two plates permit
estimating the unit cell translational velocity – here assumed equal to the bubble nose velocity – knowing that the plates
are separated by a distance of 50 mm and measuring the delay on detecting the elongated bubble front between the two
plates. Figure 4a shows the signal captured by the two plates, already transformed to the gas fraction using Eqs. (1) and
(2). A binary function is applied to the signal so as to detect the elongated bubble and the slug regions (Bertola, 2003):
0 ; if RG  RG ,crit
u RG ,t   
1 ; if RG > RG ,crit
(3)
being RG,crit a critical gas fraction above which an elongated bubble is assumed as passing through the cross section.
This critical gas fraction depends on the gas and liquid superficial velocities and is shown in Figure 4a by a dashed line.
Receiver
Transmitter
Wires
Elongated bubble
Film
hL
Flow direction
(a)
(b)
(c)
Figure 3. (a) Printed circuit board of the resistive sensor with the transmitter and receiver wires. (b) Cross sectional area
during the passage of an elongated bubble, assuming a flat interface between the phases. (c) Set of three resistivity
sensors for measuring the unit cell translational velocity.
IV Journeys in Multiphase Flows (JEM 2017)
0.6
TB
TB
TS
Gas fraction [-]
0.5
0.4
0.3
0.2
0.1
RG ,crit
0
1.665
1.67
1.675
Time (s)
(a)
1.68
Time (s)
(b)
Figure 4. Gas fraction from two resistivity sensors separated by 50 mm: (a) as measured and (b) after applying the
binary function of Eq. (3) to recognize the elongated bubble and slug regions.
After applying Eq. (3) over the resistive sensor signal, it becomes a squared wave, as shown in Figure 4b. The
period for an elongated bubble and for a liquid slug to pass through the resistive sensor are indicated by TB and TS,
respectively. The delay in capturing the elongated bubble front between the two distinct plates is named TB . Since the
distance between the two resistive sensors dS is known, the elongated bubble translational velocity can be estimated as
shown in Eq. (4). The elongated bubble and the slug lengths can be estimated assuming that the entire unit cell
translates with the same velocity of the elongated bubble front UT (Taitel and Barnea, 1990), as presented in Eq. (5).
The slug flow frequency is estimated as the inverse of the unit cell period of passage, Eq. (6).
U T  d S TB
LB  U T TB ; LS  U T TS
(4)
(5)
f  TB  TS 
(6)
1
Near each set of resistive sensors, a gauge pressure transducer is positioned to estimate the local gas superficial
velocity, which is accomplished by comparing the gauge pressures at the test section inlet and at the orifice plate. Using
the volumetric gas flow rate measured by the orifice plate QG,orifice plate and dividing by the cross sectional area of the
pipe:
jG,test section =
Porifice plate 4Q G,orifice plate
Ptest section
 D2
(7)
3. RESULTS AND DISCUSSIONS
Before starting the analysis of the results, it is interesting to understand the main aspects brought by the introduction
of the solid particles in the flow. These aspects can be divided into three groups:
(A) Increase in the mixture superficial velocity: the fixed parameters during the measures were the liquid (jL) and
gas (jG) superficial velocities. However, the presence of a slurry flow is associated to the insertion of a solid
superficial velocity (jS). Shoham (2006) shows that, for an ideal case of non-slip flow, the volumetric fraction
of the particles inside the liquid can be estimated as the ratio between the solid and the liquid superficial
velocities, RS / L  jS jL . This is a reasonable assumption for homogeneous solid-liquid flows and for small
differences between the densities of the phases, which is, indeed, the case of the present study. Upon isolating
the solid superficial velocity, it can be written as jS  RS / L jL . Shoham (2006) also affirms that the mixture
superficial velocity is the sum of the superficial velocities of all the phases, J  jL  jG  jS . Since the gas and
liquid superficial velocities remained constant, but the solid concentration changed, then the mixture
superficial velocity increases in J  jS  RS / L jL .
(B) Properties variation: the homogeneous solid-liquid dispersion can be treated as a liquid phase with equivalent
properties, dependent on the particles concentration. That is, the three-phase solid-liquid-gas flow can be
L.M.M. Rosas, C.L. Bassani, M.A. Marcelino Neto, M.J. da Silva, R.E.M. Morales, A.K. Sum
Experimental Analysis of Horizontal Three-Phase Solid-Liquid-Gas Slug Flows with Hydrate-Like Particles
treated as a two-phase dispersion-gas flow. The dispersion density can be calculated via an homogeneous
model (Shoham, 2006) as  L   S RS / L   L 1  RS / L  . The notation ( ) L represents the dispersion, being the
index L an indication that it can be treated as homogeneous liquid phase. The dispersion viscosity can be
estimated by Krieger and Dougherty’s (1959) correlation for spherical particles with maximum package factor
1.575
.
of 0.63,  L   L 1  RS / L 0.63
(C) Particles interference on the liquid velocity profile: the particles are submitted to different forces, such as
weight, buoyancy, drag and lift (Peker and Helvaci, 2007). Depending on the magnitude of those forces,
movement may be generated between a single particle and the liquid continuous phase. Even though in the
average all particles are carried out by the liquid at approximately the same velocity – as previously assumed
that no-slippage occurs – locally there is a relative movement between the particle and the liquid phase. The
drag induced by the liquid over the particle results in a terminal velocity Vt. If the particle Reynolds number is
sufficiently high ( ReP   LVt d P  L  100) (Crowe et al., 2012), being dP the particle diameter, then vortices
will be generated at the particle’s wake zone, which by their turn will disturb the liquid velocity profile.
Next, the results will be discussed in means of the aforementioned effects brought by particles introduction in the
gas-liquid slug flow. The results will be divided in two subsections: a qualitative analysis, based on the High Speed
Camera results; and a quantitative analysis, based on the resistive sensors results.
3.1. Flow visualization
Figure 5 presents an assembly of different images along the same elongated bubble. The chosen pair of superficial
velocities for this analysis is jG = 1 m/s and jL = 0.5 m/s. The two particle concentration cases are shown together with
the case of gas and liquid flow only. It is important to notice that, for all measured data, the solid particles stayed
homogeneously dispersed in the liquid. Heterogeneous distributions and moving and/or stationary beds are not
presented in this study.
The presence of 8 g/dm3 ( RS / L  0.0085) of solid particles in the jG = 1 m/s and jL = 0.5 m/s flow generated: (i) a
mixture superficial velocity variation of J  0.008 m s (approximately 0.5% of mixture acceleration); (ii) a dispersion
density of  L  994 kg m3 (approximately 0.1% smaller than the water density); (iii) a dispersion viscosity of
 L  8.9  104 Pa.s (approximately 2% higher than the water viscosity); and (iv) a particle Reynolds number of
ReP  2.5 – using Zigrang and Sylvester (1981) correlation for the terminal velocity.
(a) Without solid particles
Slug
Flow direction
Elongated bubble
Bubble nose
Film
Wake zone
3
(b) Solid particle concentration: 6 g/dm
3
(c) Solid particle concentration: 8 g/dm
Figure 5. Image assembly of different parts of the elongated bubble for jG = 1 m/s and jL = 0.5 m/s for particles
concentrations of: (a) 0 g/dm3, (b) 6 g/dm3 and (c) 8 g/dm3.
From Figure 5, it can be noticed that the elongated bubble has an aerodynamic shape, and its nose travels a little bit
displaced from the upper wall. This displacement is slightly pronounced when the solid particles are present. Taitel and
Barnea (1990) discuss the elongated bubble shape in means of a gas and liquid combined momentum equation. By their
theory, the density variation due to the introduction of the solid particles – effect (B) – would affect the inertial terms of
IV Journeys in Multiphase Flows (JEM 2017)
the momentum balance. In the same way, the viscosity variation would change the friction terms – also effect (B).
Furthermore, the liquid velocity profile at the rear of the slug body is related to the elongated bubble nose shape due to
local shear stresses, and the variation in the liquid velocity profile due to the introduction of the solid particles – effect
(C) – could be another explanation for the nose shape change. However, the particle Reynolds number seems too low
for effect (C) to be present.
Figure 6 presents a similar assembly of images, but for the slug region. An increase in the dispersed bubbles
released in the elongated bubble tail can be noticed when introducing the solid particles. This region, known as bubble
wake, presents a high liquid recirculation due to the shock of two different velocities: the slug and the film ones.
Usually, the slug travels faster than the film, which results in a liquid mass exchange at the bubble rear, phenomenon
known as scooping (Shoham, 2006). The liquid captured by the slug causes a local change on the velocity field, with a
consequent recirculation zone and a hydraulic jump. The introduction of the particulate material will intensify this
recirculation due to the interaction of the particles with the liquid velocity profile – effect (C) – if the particle Reynolds
number is sufficiently high, thus causing higher shear stresses and higher amount of gas dispersed bubbles release –
which is indeed observed in Fig. 6. The particle Reynolds number, as evaluated in the present work, may not be
representative for the elongated bubble wake zone, due to the characteristic recirculation at this region and since ReP as
evaluated in the present work does not take into account this case. Higher particle Reynolds numbers are expected to be
present at this region; however, it is hard to establish if they are sufficiently high to be a cause of the higher amount of
dispersed bubbles release.
Some of the dispersed bubbles recirculate, reaching again the elongated bubble tail and re-coalescing. Others just
travel in the wake zone, with a velocity approximately equal to the elongated bubble rear. However, some of the
dispersed bubbles detach from the wake zone, reaching the slug body. The slug body does not present high recirculation
as in the wake zone, and thus the dispersed bubbles tend to agglomerate in the upper wall due to buoyancy forces. Since
these bubbles are near the wall, they travel at a smaller velocity in comparison to the liquid in the central part of the slug
body. As a consequence, they are left behind to be captured by the next elongated bubble.
Once the dispersed bubbles reach the tail of the slug body, they can either: (i) coalesce to the next elongated bubble
or (ii) accumulate at the upper part of the elongated bubble nose. Mechanism (i) was observed as being predominant in
the absence of the solid particles, while mechanism (ii) was predominant when the particles were introduced. This
points out that the solid particles may retard bubble coalescence, which by its turn can be related to: (i) surface tension
variation – effect (B); or (ii) capillary interaction between particle and gas-liquid interface. Further analysis should be
made to understand which is causing the bubble coalescence to retard – if surface tension or capillary forces – and
closer images of the particles and interface interaction will probably be required.
(a) Without solid particles
Wake zone
Dispersed bubble coaslescing with elongated one
3
(b) Solid particle concentration: 6 g/dm
Dispersed bubble accumulation
3
(c) Solid particle concentration: 8 g/dm
Figure 6. Image assembly of different parts of the slug region for jG = 1 m/s and jL = 0.5 m/s for particles concentrations
of: (a) 0 g/dm3, (b) 6 g/dm3 and (c) 8 g/dm3.
3.2. Effects of the solid particles in the slug flow parameters
In this subsection, the results obtained with the resistivity sensors will be shown as to understand the effects of the
solid particles presence in the main slug flow parameters. The slug flow parameters analyzed are: the unit cell
translational velocity, the slug flow frequency and the elongated bubble and slug lengths. The left column of Figure 7
presents the average results of these parameters in means of the solid particles concentration for all the gas-liquid
L.M.M. Rosas, C.L. Bassani, M.A. Marcelino Neto, M.J. da Silva, R.E.M. Morales, A.K. Sum
Experimental Analysis of Horizontal Three-Phase Solid-Liquid-Gas Slug Flows with Hydrate-Like Particles
jG = 0.25 ; jL = 0.75 [m/s]
jG = 0.50 ; jL = 0.50 [m/s]
jG = 0.50 ; jL = 1.00 [m/s]
(a) 2.5
jG = 0.75 ; jL = 0.75 [m/s]
jG = 1.00 ; jL = 0.50 [m/s]
jG = 0.50 ; jL = 1.50 [m/s]
8
jG = 1.00 ; jL = 1.00 [m/s]
jG = 1.50 ; jL = 0.50 [m/s]
0 g/dm3
6 g/dm3
8 g/dm3
6
PDF
UT [m/s]
2
4
1.5
2
1
0
0
(b) 12
2
4
6
Particles concentration [g/dm3]
8
1.4
1.6
1.8
UT [m/s]
2
2
2.2
1.5
8
PDF
Frequency [Hz]
10
6
1
4
0.5
2
0
0
0
2
4
6
Particles concentration [g/dm3]
8
(c) 2.5
0
0.5
1
1.5
2
Frequency [Hz]
2.5
3
0
0.5
1
2.5
3
1.5
1
1.5
PDF
LB [m]
2
1
0.5
0.5
0
0
0
2
4
6
Particles concentration [g/dm3]
8
0.4
4
0.3
3
PDF
5
LS [m]
(d) 0.5
0.2
0.1
1.5
LB [m]
2
2
1
0
0
0
2
4
6
Particles concentration [g/dm3]
8
0
0.2
0.4
0.6
0.8
1
LS [m]
Figure 7. Results obtained with the resistive sensor positioned at 250D from the pipe inlet for: (a) unit cell translational
velocity, (b) slug flow frequency, (c) elongated bubble length and (d) slug length. Left column: average results for all
gas-liquid superficial velocities in means of the solid particles concentration. Right column: Probability Density
Function (PDF) for the different particles concentration when jG = 1 m/s and jL = 0.5 m/s.
IV Journeys in Multiphase Flows (JEM 2017)
superficial velocities. The right column shows the Probability Density Function (PDF) for the different solid particles
concentration for fixed gas and liquid superficial velocities of jG = 1 m/s and jL = 0.5 m/s.
Figure 7a presents the results obtained for the unit cell translational velocity. It can be observed that the average
value of the unit cell increases as the solid particles are introduced in the flow, Figure 7a (left). This unit cell
acceleration may be caused by the increase in the mixture superficial velocity due to the introduction of the solid
particles – effect (A). From Figure 7a (right), it can be seen that the peak of the PDF displaces to the right as the solid
particles are introduced, indicating an average acceleration of the elongated bubble. However, the unit cell velocity
distribution is larger for higher particles concentration. That is, in the average, the solid particles introduction
accelerates the elongated bubble, but some elongated bubbles may present smaller values of translational velocity due
to the intermittence of the flow. For example, the lower value of velocity for 8 g/dm3 found was near 1.45 m/s, while for
the case of none solid particles was approximately 1.55 m/s.
Figure 7b presents the results for the slug flow frequency. The average value of the slug flow frequency tends to
increase with the introduction of the solid particles, as shown in Figure 7b (left). This is a consequence of the unit cell
acceleration, which causes the unit cell structures to have lower periods of passage. The PDF is slightly displaced to the
right when the solid particles are introduced. The probability peaks are lower, showing a higher dispersion of the
frequency results in the presence of the particles.
Figures 7c,d present the results for the elongated bubble and slug region lengths. It is noticeable that their average
values tend to decreases (Figures 7c,d – left). When analyzing the PDF’s, the elongated bubble (Figure 7c – right)
presents a peak value reduction from approximately 1.3 m to 1 m when the solid particles are introduced. In the other
hand, the variation on the PDF of the slug length (Figure 7d – right) is less sensitive to the particles introduction, being
the peak displaced from approximately 0.35 m to 0.28 m. The region lengths are strictly related to the elongated bubble
shape (Bassani et al., 2016; Taitel and Barnea, 1990), and since the properties variations due to the formation of a
dispersion – effect (B) – affect the bubble shape, than it would also affect the region lengths.
4. CONCLUSIONS
This study presents an experimental analysis of three-phase solid-liquid-gas slug flows using particles with similar
properties of gas hydrates. High Speed Imaging and resistive sensors techniques were used to characterize the slug flow
pattern in the presence of solid particles. The main effects introduced by the solid particles can be split into: (A) a
mixture superficial velocity increase due to the particles superficial velocity, (B) properties variations due to the
formation of a dispersion and (C) the induction of vortices in the wake of the particles, which will disturb the liquid
velocity profile. Effect (A) is related to the unit cell acceleration, with consequent slug flow frequency increase. Effects
(B) is related to variations on the elongated bubble shape, especially at the elongated bubble nose, which tends to travel
more displaced from the upper wall at the presence of the particles. Effect (C) may be related to higher shear stresses on
the elongated bubble wake zone, thus causing a higher release of dispersed bubbles; however, it is hard to estimate the
particle Reynolds number at this region and therefore further analysis should be made on verifying cause-effect relation.
5. ACKNOWLEDGEMENTS
The authors acknowledge the financial support of ANP and FINEP through the Human Resources Program for Oil
and Gas Segment PRHANP (PRH 10-UTFPR), TE/CENPES/PETROBRAS (0050.0068718.11.9) and National Council
for Scientific and Technological Development (CNPq). AKS thanks PETROBRAS for sponsoring his sabbatical leave
at UTFPR during the time part of this study was performed.
6. REFERENCES
Bassani, C.L., Pereira, F.H.G., Barbuto, F.A.A., Morales, R.E.M., 2016. Modeling the scooping phenomenon for the
heat
transfer
in
liquid-gas
horizontal
slug
flows.
Appl.
Therm.
Eng.
98,
862–871.
doi:10.1016/j.applthermaleng.2015.12.104
Bertola, V., 2003. Experimental characterization of gas-liquid intermittent subregimes by phase density function
measurements. Exp. Fluids 34, 122–129. doi:10.1007/s00348-002-0543-3
Camargo, R.M.T., Palermo, T., Sinquin, A., Glenat, P., 2000. Rheological Characterization of Hydrate Suspensions in
Oil Dominated Systems. Ann. N. Y. Acad. Sci. 912, 906–916. doi:doi:10.1111/j.1749-6632.2000.tb06844.x
Cardoso, C.A.B.R., Gonçalves, M.A.L., Camargo, R.M.T., 2015. Design options for avoiding hydrates in deep offshore
production. J. Chem. Eng. Data 60, 330–335. doi:dx.doi.org/10.1021/je500601f
Castillo, F.V., 2013. Caracterização experimental do escoamento intermitente líquido-gás em tubulações horizontais.
Dissertação de Mestrado, Universidade Tecnológica Federal do Paraná, Curitiba/PR, Brasil.
Crowe, C.T., Schwarzkopf, J.D., Sommerfeld, M., Tsuji, Y., 2012. Multiphase Flows with Droplets and Particles, 2nd
ed, CRC Press New York.
Doron, P., Barnea, D., 1993. A three-layer model for solid-liquid flow in horizontal pipes. Int. J. Multiph. Flow 19,
L.M.M. Rosas, C.L. Bassani, M.A. Marcelino Neto, M.J. da Silva, R.E.M. Morales, A.K. Sum
Experimental Analysis of Horizontal Three-Phase Solid-Liquid-Gas Slug Flows with Hydrate-Like Particles
1029–1043. doi:10.1016/0301-9322(93)90076-7
Doron, P., Granica, D., Barnea, D., 1987. Slurry flow in horizontal pipes-experimental and modeling. Int. J. Multiph.
Flow 13, 535–547. doi:10.1016/0301-9322(87)90020-6
Fox, R.W., Pritchard, P.J., McDonald, A.T., 2011. Fox and McDonald’s Introduction to Fluid Mechanics, 8th ed. Sons,
John Wiley &, Hoboken, USA.
Goharzadeh, A., Rodgers, P., 2009. Experimental characterization of solid particle transport by slug flow using Particle
Image Velocimetry. J. Phys. Conf. Ser. 147, 12069.
Goharzadeh, A., Rodgers, P., Touati, C., 2010. Influence of gas-liquid two-phase intermittent flow on hydraulic sand
dune migration in horizontal pipelines. J. Fluids Eng. Trans. ASME 132, 713011–713017. doi:10.1115/1.4001869.
Kjeldby, T.K., Henkes, R. a W.M., Nydal, O.J., 2013. Lagrangian slug flow modeling and sensitivity on hydrodynamic
slug initiation methods in a severe slugging case. Int. J. Multiph. Flow 53, 29–39.
doi:10.1016/j.ijmultiphaseflow.2013.01.002
Krieger, I.M., Dougherty, T.J., 1959. A mechanism for non-newtonian flow in suspensions of rigid spheres. Trans. Soc.
Rheol. 3, 137–152.
Machado, D., Hildebrando, F., Lipinski, L., 2013. Sistema Multicanal de Sensores Resistivos para Medição de
Escoamentos Bifásicos. Universidade Tecnológica Federal do Paraná.
Peker, S., Helvaci, S., 2007. Solid-liquid two-phase flow, 1st ed. Elsevier Science, Amsterdam, Netherlands.
Rogero, E.C., 2009. Experimental Investigation of Developing Plug and Slug Flows. Tecnhische Universität München.
Shoham, O., 2006. Mechanistic modeling of gas-liquid two-phase flow in pipes, 1st ed. Society of Petroleum Engineers,
Richardson, USA.
Sloan, D., Koh, C., Sum, A.K., 2011. Natural gas hydrates in flow assurance, 1st ed. Elsevier Inc., Burlington, USA.
Sloan, E.D., Koh, C.A., 2008. Clathrate hydrates of natural gases, 3rd ed. Taylor & Francis Group, Boca Raton, USA.
Stevenson, P., Thorpe, R.B., 2003. Energy dissipation at the slug nose and the modeling of solids transport in
intermittent flow. Can. J. Chem. Eng. 81, 271–278.
Stevenson, P., Thorpe, R.B., 2002. Velocity of isolated particles along a pipe in stratified gas-liquid flow. AIChE J. 48,
963–969. doi:10.1002/aic.690480506
Taitel, Y., Barnea, D., 1990. A consistent approach for calculating pressure drop in inclined slug flow. Chem. Eng. Sci.
45, 1199–1206. doi:10.1016/0009-2509(90)87113-7
Zigrang, D., Sylvester, N., 1981. An explicit equation for particle settling velocities in solid–liquid systems. AIChE J. 6,
1043–1044.
7. RESPONSIBILITY NOTICE
The authors are the only responsible for the printed material included in this paper.