You are on page 1of 9

Turbulence, Heat and Mass Transfer 8

c
2015
Begell House, Inc.

Statistics of slug flow subjected to wall transpiration


F. J. S. Bandeira, J. B. R. Loureiro and A. P. Silva Freire
Mechanical Engineering Program (PEM/COPPE/UFRJ)
C.P. 68503,Rio de Janeiro, 21941-972, Brazil.
jbrloureiro@mecanica.coppe.ufrj.br
Abstract The purpose of the present work is to investigate the changes on the statistical properties of horizontal
slug flows caused by flow injection at the pipe wall. Measurements of turbulent flow in a horizontal slug flow
subjected to wall transpiration are presented. Results include data on global flow rates and pressure drop, and on
local mean and fluctuating velocity profiles. The properties of the two-phase flow are measured through a Shadow
++
Sizer system and laser-based sensors. Two distinct flow transpiration rates are studied, vw
= vw /Um = 0.0005 and
0.001. The effects of flow transpiration seem to lead to bubble break-up and to changes in the slug frequency. In
addition to the two-phase flow results, single-phase flow measurements are presented with a view to compare the
different turbulent effects introduced by the second phase.

1. Introduction
Turbulent flow in pipes with porous surfaces is a problem of great industrial interest. Typical applications include the thermal protection of walls, filtration or the production of oil in horizontal
or vertical wells.
For laminar flow, the Navier-Stokes equations can be solved to unveil the dependence of
the velocity components and the pressure on position coordinates, pipe dimensions and fluid
properties. Berman (1953) proposed a classical solution for two-dimensional, incompressible,
steady-flow in a channel. Other authors have examined Bermans solution to consider pressuredependent wall suction (Haldenwang, 2007) or the effect of slip boundary condition (Chellam
and Liu, 2006).
The above mentioned analyses consider symmetric flows in symmetric ducts. Extension of
the solutions to circular pipes were provided, e.g., by Erdogan and Imrak (2005) and Tsangaris
et al. (2007).
For high Reynolds number flow, the search for analytical solutions in much complicated by
the natural requirement of turbulence closure. Additionally, turbulent flow is known to be sensitive to wall roughness. For external flow, some authors have resorted (Stevenson 1963, Simpson
1970, Silva Freire 1988) to simple algebraic closure hypotheses and perturbation arguments to
develop local analytical solutions for the fully turbulent region. These solutions exhibit bilogarithmic terms and incorporate the effects of local Reynolds number and transpiration rate.
One prominent feature is the dependence of an integration parameter on the transpiration rate.
The underlying assumption of all the analytical methods is that flow transpiration at the
wall is homogeneous. Some authors, however, have approached to problem differently. They
consider distributed fluid injection, that is, isolated perforations, so that frictional losses are
described through a decomposition of effects: wall friction, perforation friction and mixing effects. Perforation friction is normally associated with an increase in roughness. The mixing
effects are compared to the problem of multiple interacting jets in a cross flow. This approach
results in such an intricate analysis that only external empirical evidence can be used to determine the correct behavior of the friction contribution.
Limited experimental works have been made for turbulent flow in porous pipes. The investigation of Olson and Eckert (1966) described the effects of continuous mass injection at the

Turbulence, Heat and Mass Transfer 8

wall in the shape of the velocity profile and in the pressure loss. Results on friction factor and
eddy diffusivity were also discussed. Su and Gudmundsson (1998) and Ouyang et al. (1998)
provided further experimental data; the latter authors, in particular, considered also two-phase
flows.
The purpose of the present work is to carry out reference experiments in horizontal pipes
with flow injection at the wall for single- and two-phase (slug) flow patterns. Measurements
of global flow rates and pressure drop are made for two different injection rates. In addition,
the work presents local mean and fluctuating turbulent profiles obtained through Laser-Doppler
Anemometry. The properties of the two-phase flow are measured through a Shadow Sizer
system and laser-based sensors. The validity of the law of resistance previously advanced by
Loureiro and Silva Freire (2011) is investigated for both the single-phase and slug flows.
Important properties of slug flow including the passage frequency of unit cells (t ) and the
length (`) distribution of long bubbles have not been previously discussed in the literature. The
flow acceleration provoked by the wall injection of fluid and the very high levels of turbulence
observed in the near wall region are bound to exert a large influence on both t and `. These
two aspects of the problem are reported here.

2. Theoretical background
The law of resistance introduced in Loureiro and Silva Freire (2011) is valid for single-phase
flow and incorporates the effects of roughness and wall transpiration. The law is derived from a
perturbation solution (see also Silva Freire (1988)); the main details of the derivation are shown
next.
2.1. Resistance law for smooth pipes
The friction coefficient can be defined by
2
(p1 p2 )L1 = (D1 )(21 )Um
,

(1)

where D denotes the pipe diameter, Um is the mean flow velocity and L the length of a fluid
cylinder.
For very large Reynolds numbers, the velocity distribution in the near wall fully turbulent
region is given by

1
uu1
ln yu 1 + A,
(2)
=
p
where y is the wall distance, u is the friction velocity (= w /), (= 0.4) is von Karmans
constant and A = 5.5.
An integration of Eq. (16) over the cross-sectional area the pipe together with algebraic
manipulations gives the resistance law for smooth pipes,
 

( )1 = 2.035 log Re 0.8,


(3)
with Re = Um D 1 .
2.2. Resistance law for smooth pipes with wall transpiration
For flow subject to wall transpiration, Eq. (3) is clearly not valid anymore. The result of
the injection or suction of fluid into an oncoming flow is to modify the velocity distribution
throughout the boundary layer so that drag is either reduced or increased. Any expression

F. J. S. Bandeira, J.B.R. Loureiro and A. P. Silva Freire.

advanced with the purpose of determining the friction coefficient should therefore reflect this.
Furthermore, an explicit dependence of Eq. (3) on the transpiration rate should be expected.
In Silva Freire (1988) the matched asymptotic expansions method was applied to the equations of motion to find a law of the wall in which the additive parameter A varied with transpiration. The resulting expression is


2


1 W y 1 ,
u+ = 1 ln y + + A + 1 W y 1 + vw+ (2)1 ln y + + 21 A +
(4)
1
+
1
+
+
1
with u = uu , y = yu , vw = vw u , vw = normal velocity at the wall and A is given by:
A = 5 512(vw U 1 ),

(5)

and function W are related to the universal wake function.


and parameters and
An integration of Eq. (4) over the cross-sectional area of a pipe, gives
2

Um = U 3.75u vw (1.86A + 2.34 ln Re+ 5.47),

(6)

with Re+ = Ru 1 .
Some further algebraic manipulations with

Um u 1 = 2 2( )1 ,

(7)

give
1=

1
(2 2) (2, 5 ln(Re+ ) + A 3, 75)

+ vw++ (1, 56 ln2 (Re+ ) + (1, 25A 4, 68) ln(Re+ ) + 41 A2 + 1, 86A + 5, 47),

(8)

where
vw++ = vw Um 1

1
Re+ = (Um D 1 )(4 2
).

and

(9)

The transcedental equation, Eq. (8), gives for given Re+ and vw++ .
2.3. Resistance law for rough pipes
The law of resistance for flow in a rough pipe was shown by Nikuradse 1933 to have the form

uu 1 = 1 ln yks 1 + B,

(10)

where ks is a characteristic length of the roughness and B = 8.5 (completely rough regime).
In fact, B was shown to be a function of Rek (= ks u 1 ). The behaviour of B for the three
types of flow regime discussed by Nikuradse 1933 has been studied by several authors. For
example, Ligrani & Moffat 1986 suggest the following functional dependence
B = 8.5 + (1 ) 1 ln (Rek ) + (1 )C,

(11)

where Rek = ks u /, C = 5.1 and = sin((1/2)g) with


g = ln Rek Re1
k,s

ln Rek,r Re1
k,s

1

Rek,s = 5, Rek,r = 70 and this approximation is valid in 5 Rek 70.

(12)

Turbulence, Heat and Mass Transfer 8

The resistance formula for flow in a rough pipe can be obtained by integrating Eq. (10) over
the cross-sectional area of a pipe. The result is
= [0.88 ln(Rks1 ) + 0.35B 1.33]2 .

(13)

A comparison of Eq. (13) with the experiments of Nikuradse shows that for a fully rough
regime the ln additive term should be replaced by 1.74.
2.4. Resistance law for rough pipes with wall transpiration
A law of resistance for rough pipes with wall transpiration can now be deduced provided the
results of the previous sections are taken into account. Let us define Re+ = R/ks and Ak = B 512 vw++ . It follows immediately from Eqs. (8) and (13) that

1 = (2 2)1 (2.5 ln(Rks1 ) + Ak 3.75) + vw++ (1.56 ln2 (Rks1 )
+ (1.25Ak 4.68) ln(Rks1 ) + 4 1A2k + 1.86Ak + 5.47).

(14)

3. Experiments
The experiments were conducted in a 15 m long porous pipe (34 mm in diameter) with uniform
fluid injection. The test section consisted of three concentric stainless-steel tubes, assembled
as shown in Fig. 1. The piping system consisted of 15 one meter long segments, all connected in line and fitted with independent pressure taps. Two plexiglass inspection windows
were installed to allow local velocity measurements and characterization of the two-phase flow
properties. Air and water were mixed in a T-junction located 1 m upstream of the inlet of the
porous pipe. A schematic diagram of the experimental setup is also shown in Fig. 1.

Figure 1: Description of the test section: a) geometric arrangement, b) photograph of actual test
section, c) schematic diagram.
Turbulent statistics were measured by a two-component Dantec Laser-Doppler Anemometry. The properties of the slug flow were measured through a Dantec Shadow Sizer System.
Additionally, laser diodes and photodetectors were used to measure the statistical properties of
the two-phase flow along the pipe, so that changes in the slug frequency, the bubble velocity
and the lengths of the liquid slug and the liquid film could be measured and spatially correlated.
The pressure losses in a porous pipe with wall injection must increase the momentum flux
and overcome wall friction. A momentum balance on a control volume of length dx and radius
R shows that the wall shear stress in given by


1
dp
d(2m )
w =
+
(15)
4 d(x/D) d(x/D)

F. J. S. Bandeira, J.B.R. Loureiro and A. P. Silva Freire.


where , the momentum flux factor, is the ratio of the real momentum flux through a given cross
section to that based on one-dimensional flow at the mean velocity.
In our estimates, was found to be 1.06.

4. Results and discussion


4.1. Single-phase flow
The roughness length (ks ) of the producing pipe was determined from the single-phase water
flow with no injection at the wall. Various experiments with Reynolds number varying from
6,000 to 100,000 were performed to find ks (= 0.000334 m). Two distinct flow transpiration
rates are studied, vw++ = vw /Um = 0.0005 and 0.001.
To investigate the logarithmic character of the velocity profile, Eq. (4) is re-written in the
form
h
i
1/2
+ 1
+ +
= 2(vw )
vw u + 1
1 = 1 ln y + + A
(16)
For the single-phase flow case, the mean velocity profile for all transpiration rates are shown
in Fig. 2a according to the coordinates defined by Eq. (16). The existence of a bi-logarithmic
region is evident, with the level of the log-region decreasing as vw increases. The agreement
provided by the predictions of Eq. (14) with the experimental data is very good despite the fact
that the parameters in Eq. (14) were not particularly adjusted to fit the present experimental
data. The values of the constants are the values presented in Loureiro and Silva Freire (2011),
based entirely in the analysis of Silva Freire (1988).

Figure 2: Comparison between the results for unblown case and for the two different transpiration rates studied: a) logarithmic mean velocity distributions, b) pressure drop distributions.

4.2. Two-phase flow


Regarding the two-phase flow experiments, mean velocity profiles for the liquid phase for two
different injection rates are shown in Fig. 3a. For the same injection velocity, the single- and
two-phase velocity profiles correspond to flows with about the same top centerline velocity.
The presence of the gas phase makes the liquid velocity profiles less full, as expected. The
injection of fluid through the wall clearly modifies the velocity distribution, reducing the wall
shear stress.
The streamwise velocity fluctuations are shown in Fig. 3b. The presence of bubbles in the
flow naturally increases its degree of agitation, with a corresponding increase in longitudinal

Turbulence, Heat and Mass Transfer 8

Figure 3: Laser-Doppler anemometry measurements: (a) mean velocity profiles and (b) longitudinal velocity fluctuations for the single and two-phase flow conditions with and without flow
transpiration. Profiles were normalized by pipe diameter D and centerline velocity U0 .
velocity fluctuations. For the highest injection rate, a peak value of 0.17U0 can be observed
near the wall. The very pronounced level of turbulence near the wall induces the observed
break-up of long bubbles and increases the level of aeration of the liquid pistons.

Figure 4: Two-phase flow characterization: (a) general pattern at the inlet of the permeable
pipe; typical noses of bubbles measured 12 m downstream of the inlet: (b) vw++ = 0.0000, (c)
vw++ = 0.0005 and (d) vw++ = 0.001.
The frictional pressure losses are shown in Fig. 5. An increase of the mixture velocity for
unblown flows increases the pressure drop. However, with positive transpiration rates, a definitive assessment could not be made. For the lower injection rate the pressure loss diminished,
whereas for the higher injection rate the pressure loss increased. The two-phase flow problem
in fundamentally different from the single-phase flow problem in two aspects: the passage frequency of unit cells and the length of the large bubbles are strongly influenced by the injection
rate.
The general two-phase flow pattern upstream of the inlet of the permeable pipe is illustrated
in Fig. 4a. To evaluate the influence of the transpiration flow rate on the slug flow behaviour, the
Shadow Sizer System was positioned at a transparent window located 12 meters downstream
of the inlet section. A NanoSense MKIII camera provided high resolution images (1289 x

F. J. S. Bandeira, J.B.R. Loureiro and A. P. Silva Freire.

Figure 5: Pressure drop for two-phase flow with wall transpiration.

Turbulence, Heat and Mass Transfer 8

1024 pixels) at 2000 fps. Figures 4(b-d) illustrates typical noses of bubbles measured at the
transparent test section for the case of no injection (4b), for the lower injection rate, vw++ =
0.0005 (4c) and for the higher injection rate vw++ = 0.001 (4d). An increase in transpiration
flow pushes the noses of the bubbles towards the center of the pipe. The enhanced near-wall
turbulence provokes instability waves on the bubble contour that favor bubble break up and
aeration of the liquid piston.
The Shadow Sizer images and the data obtained from laser-based sensors for bubble detection were processed according with the procedure introduced in Matamoros et al. (2014)
for calculation of the statistical properties of the slug flow along the permeable pipe. Bubble
reconstruction was made through the fixed window method.
The shapes of two typical bubbles are shown in Fig. 6. The perturbations provoked by the
fluid injection increase the mean size of the bubbles and decreases their frequency passage (Fig.
7). to the present experimental conditions the increase in size was about 25%. The frequency
was reduced in 32%.

Figure 6: Bubble shape in the inlet region (top) and test section (bottom, Vw++ = 0.001).

Figure 7: Bubble frequency and length distributions in the inlet and test section regions.

5. Conclusions
The present work has studied the behaviour of single- and two-phase flows in horizontal pipes
with fluid injection at the wall.
Three different flow conditions have been scrutinized through measurements of flow rate,
pressure drop, statistics of the turbulence and characterization of the two-phase flow. The results
have shown that Eq. (14) furnishes predictions that are good with an accuracy of about 3% for
single-phase flows.

F. J. S. Bandeira, J.B.R. Loureiro and A. P. Silva Freire.

The transpiration and the air-bubble interaction were observed to promote near-wall turbulence, imposing considerable changes to the inlet flow. An increase in transpiration tends to
push bubble noses towards the center of the permeable pipe, induce bubble coalescence and
increase the level of aeration on the liquid piston.

References
1. A. S. Berman, Laminar flow in channels with porous walls, J. App. Physics, 24, 12321235, 1953.
2. S. Chellam and M. Liu, Effect of slip on existence, uniqueness, and behavior of similarity
solutions for steady incompressible laminar flow in porous tubes and channels,Physics of
Fluids, 18, 083601, 2006.
3. M. E. Erdogan and C. E. Imrak, On the axial flow of an incompressible viscous fluid in a
pipe with a porous boundary, Acta Mechanica, 178, 187197, 2005.
4. P. Haldenwang, Laminar flow in a two-dimensional plane channel with local pressuredependent crossflow, J. Fluid Mech., 593, 463473, 2007.
5. J. B. R. Loureiro and A. P. Silva Freire. Slug flow in horizontal pipes with transpiration at
the wall, Journal of Physics. Conference Series (Online), 318, 2011.
6. L. M. C. Matamoros, J. B. R. Loureiro, A. P. Silva Freire, Length-area-volume of long
bubbles in horizontal slug flow, Int. J. Multiphase Flow, 65, 2430, 2014.
7. R. Olson and E. Eckert, Experimental studies of turbulent flow in a porous circular tube
with uniform fluid injection through the tube wall, J. of Applied Mech., 33, 717, 1966.
8. L.-B. Ouyang, N. Petalas, S. Arbabi, D. E. Schroeder, K. Aziz, An experimental study of
single-phase fluid flow in horizontal wells, SPE Western Regional Meeting, Bakersfield,
California, 10 pp., 1998.
9. A.P. Silva Freire. An asymptotic solution for transpired incompressible turbulent boundary
layers, Int. J. Heat Mass Transfer 31:101l-1021, 1988.
10. R. L. Simpson, The turbulent boundary layer on a porous plate: an experimental study
of the fluid dynamics with injection and suction. Ph.D. thesis, Stanford Univ., Stanford,
California, 1967
11. T. N. Stevenson, A law of the wall for turbulent boundary layers with suction or injection.
Cranfield Rep. Aero., 166, The College of Aeronautics, Cranfield, England, U.K., 1963.
12. Z. Su and J. Gudmunsson, Perforation inflow reduces frictional pressure loss in horizontal
wellbores, J. Petroleum Sci. Eng., 19, 223232, 1998.
13. S. Tsangaris, D. Kondaxakis, N.W. Vlachakis, Exact solution for flow in a porous pipe
with unsteady wall suction and/or injection, Comm. Nonlinear Sci. Num. Simulation, 12,
11811189, 2007.