Experimental Study of the Pipeline Lubrication for Heavy Oil Transport
Oil & Gas Science and Technology - Rev. IFP, Vol.
Experimental Study of the Pipeline Lubrication for Heavy Oil Transport
A. Bensakhria 1
Y. Peysson 0
G. Antonini 1
0 Institut français du pétrole , 1 et 4, avenue de Bois-Préau, 92852 Rueil-Malmaison Cedex - France
1 Université de technologie de Compiègne, Centre de recherches de Royallieu, UMR 6067 du CNRS , BP 20529, 60205 Compiègne Cedex - France
- Experimental Study of the Pipeline Lubrication for Heavy Oil Transport - Heavy oil is accessible in different areas around the world in large amount. Unfortunately its high viscosity makes it difficult to produce and to transport. In this article we focus on transport problems. Pressure drop in the pipes must be lower as possible to limit pump power and to be able to transport in long distance. In the case of heavy oil, the high viscosity leads to huge pressure drop that makes it impossible to simply pump the fluid in single-phase flow, even of large diameter. Different solutions have been developed to transport heavy oil in pipeline like dilution of the crude oil in a lighter one. In this study we experiment a solution that do not modify the viscosity of the oil phase but transform the flow regime of transport: pipeline lubrication. Pipeline lubrication is a technique based on core-annular
flow regime and used for pressure drop reduction in the transport of very viscous products. A thin water
film adjacent to the internal pipe wall lubricates the internal oil core leading to a longitudinal pressure
An experimental study has been carried out where the influence of the different parameters, which could
affect the lubrication process, has been optimised, using water as the annulus and heavy oil as the core of
the flow. The tests were conducted in steady laminar flow at moderate flow rates. The results, obtained
with annular water injection at the pipe wall, show a pressure drop reduction over than 90% as
compared with the same product in flow without lubrication. These results confirm the effectiveness of
the lubricating process for heavy oil transport. We also showed that density difference is responsible to
particular evolution of the pressure drop with the flow rate.
The interest in heavy oil production increased in the recent
years because of the large amount of reserve accessible.
Estimation of the world reserve is difficult, but the order of
magnitude of the total volume of heavy oil is the same that
conventional oil. But, a major difficulty with heavy oil is its
huge viscosity that makes it challenging to extract. A key
issue to assure the production and the transportation of the
hydrocarbon phase is to be able to decrease the viscosity.
Indeed, mobility is needed in the porous media. The Darcy
law shows that for high viscosity, the flow rate is very weak
and exploitation is not economical. Enhancement is required
for heavy oil production. Hot vapour injection or in situ
combustion has been proposed to decrease viscosity in the
field by heating. But pipe transportation is also difficult
because of the high viscosity.
In this study, we focus on transport properties through
pipelines. The simple single-phase flow of heavy oil in a tube
will lead to huge pressure drop that makes it impossible to
pump. Different solutions are then proposed to allow the
flow. Dilution of the heavy oil in a lighter phase is the most
common way to reduce friction. Light oil is then needed in
very large quantities and recycling is built for industrial
development. Light oil is separated at the treatment plant and
a second parallel pipe transports it again at the field area.
However, the need of light oil is a large limitation of this
In this study, we investigated experimentally a transport
solution by Core Annular Flow (CAF). CAF is one particular
flow regime where the oil phase is in the centre of the pipe
and water is flowing near the wall surface. A very pleasant
characteristic of this flow is that the pressure drop is close to
that of pure water and does depend only weakly on the heavy
This remarkable property has been observed for long and
its industrial interest was noticed a hundred years ago! A
1904 patent of Isaacs and Speed  in the United States
mentioned first the ability to transport viscous product
through “water lubrication”. Despite this early concern, a
large-scale industrial pipeline for heavy oil was built only in
the 1970’s. This Shell line near Bakersflield in California was
38 km long for a tube diameter of 15 cm. For more than ten
years, a viscous crude oil has been produced at the flow rate
of 24 000 bbl/d in water lubricated regime.
Since then, several studies were dedicated to CAF regime,
and different reviews of the published work have been
written [2, 3].
1 FLOW REGIME
Several flow regime configurations are possible when
flowing a mixture of water and oil in a tube as shown in
Figure 1. Multiphase flows of liquid-liquid, gas-liquid are
crucial in the oil and gas industry and a large amount of
work is available on flow regime characterisation. A
complete overview on multiphase flow has been made in the
1970’s , and two-fluid dynamics are synthesised more
The Figure 1 represents various flow regimes when oil
and water are flowing for different relative quantity of the
two phases. The configurations are depending on the fluid
properties such as density, surface tension and also on the
shear rate in the flow. Mean injection velocities are also key
parameters for the flow regime determination.
For fixed injection flow rate, the Figure 1 represents
different configurations obtained when varying the relative
amount of oil in water. Emulsion of small droplets is found
for few quantity of oil in water. If more oil is added, the
droplets are growing in size and became the order of
magnitude of the pipe radius. Oil slugs appear in the water
phase. With the increase of oil in the system, slugs can merge
and the oil phase became continue on the pipe length.
Stratified flow and CAF regime are observed in those
conditions. When the oil fraction become close to one, we
get a continuous oil phase with water droplets.
Core annular flow is one of the flow regimes observed in
two-phase flow. But a very specific property is that the
pressure drop for the system is the lowest one amount all the other
flow regime for the equivalent liquid and water flow rate.
In this flow regime, water is at the pipe surface and
lubricates the oil core. Perfect CAF is shown in Figure 2. A
small water layer is sheared and the velocity field is
Different flow regimes of two-phase flow (oil in water). From left to right, the relative volume of oil in water is increasing .
Many authors investigated the question of the stability of
such system. Because of its simplicity, the perfect CAF
without density difference was first analysed. It has been
shown that stability in this case is achieved for only a small
range of parameters . For a fixed volume ratio between oil
and water, the core annular flow is not stable at low velocity.
The capillary instability due to surface tension appears and
breaks the core. But the increase of velocity stabilised the
capillary instability and the flow regime can then be
observed. But for larger velocity, PCAF flow become again
unstable due to the interfacial friction, undulation of the flow
arises leading to WCAF. But this configuration can stabilise
again. Only much higher velocities can give rise to break up
of the core or emulsification process.
The velocity range of existence of CAF is represented in
Figure 3 in the case of experimental work reported by .
The particular range of stability raised two main problems
for industrial development. First, the operating conditions
(pump power, pipe geometry, etc.) must be compatible with
the stability of the flow, and second, if CAF is not stable at
low velocity, the stop and restart of the flow must be
3 PRESSURE DROP
Despite the problems mentioned above, a huge advantage of
CAF is that the pressure drop is the smallest one of the
twophase flow regime. The transport of very viscous crude oil is
then possible with pressure drop of water! Guevara et al. 
Velocity r R
Rs µ w(r)
Flow of oil in water in a perfect core annular flow regime.
Schematic presentation of the velocity and the viscosity
approximately linear if the difference of viscosity between oil
and water is large. In that case, the oil core is nearly a plug
flow. Very weak deformations take place.
Perfect CAF appears to be very rare and can exist only for
density matched fluid. Several experimental observation have
shown that waves are creating at the water and oil interface
 leading to Wavy Core Annular Flow (WCAF). This type
or CAF seems to be the one observed in real situation.
Indeed, for fluids with a difference of density, a buoyancy
force will give a radial movement of the core. If no
counterbalancing force is applied, the buoyant effect will push the
core to the upper wall of the pipe. It has been shown that
waves at the interface are necessary to create sufficient
lubrication force capable of counterbalancing the buoyancy
force . Mechanisms that lead to formation of such wave at
the interface are not yet completely understood.
Stratified flow 0.01 0.01
compared different processes to transport heavy oil and
Figure 4 shows the pressure drop variation with the flow rate
for different transportation systems. CAF give rise to the
smallest pressure drop not far from pure water in the pipe.
Pressure drop measurement and friction factor estimation
have been proposed in the literature. Arney et al. [
analysed experimentally and theoretically the friction factor
variation with a specific Reynolds number in the case
of concentric CAF. Eccentricity of the core in the case
of density difference has been studied [
] and friction
factors were estimated for different position of the core in
Very low pressure gradient in the case of CAF raised very
much interest from different industry dealing with viscous
paste. In the oil industry, the heavy oil transport problem
gives rise to large concerns from the operational companies.
4 INDUSTRIAL DEVELOPMENT
We already mentioned the Shell line near Bakersfield in
California, but a line is also mentioned in the literature to
transport the bitumen produced in the area of the Orinoco
belt in Venezuela. It was a 55 km long pipeline transporting
highly viscous oil (1.5 Pa·s). This commercial line was
instrumented to study carefully the capacity of this technique
to transport heavy oil in Venezuela [
These two lines and some others detailed in different
], showed the capacity of this technique to
transport heavy oil at the industrial scale despite the different
1 000 000
Indeed, restart conditions are not in annular flow regime
and restart pressure could be high. A second observation is
the adhesion of the oil at the wall in flow conditions. The
pipe section can be restricted and increase of pressure drop
and even blockage can be observed.
Different solutions have been proposed to overcome these
difficulties (cement treatment of the pipe inner surface [
for example), however, the CAF is not a technique widely
used in the oil transport industry certainly because of the
unusual behaviour of such system.
In this study, we analysed experimentally the flow
of a very viscous heavy oil (around 5 Pa·s) and showed
the actual feasibility of CAF to transport heavy
5 EXPERIMENTAL SET-UP AND PROCEDURE
The installation used to perform the experiments consists of a
flow loop (Fig. 5) containing the heavy oil storage tank and a
stainless steel pipe of 12 m total length of 25 mm internal
diameter, equips with a fluid lubricant injector, especially
designed for pipe-wall lubrication and a data acquisition
The heavy oil, stored in a tank of 450 l capacity, is
transferred using a screw-type positive displacement
(moyno) pump to the center of the injector (Fig. 6) located at
the head of the pipe. The flow rate is controlled by the
variation of the speed of the pump. Water (fluid of
lubrication) is drown from a supply stainless steel tank of 50 l
and pumped into the annulus of the injector using a
volumetric pump, assuring the intern pipe-wall lubrication.
Data acquisition system
Pressure drops within the pipeline is measured across
a test section of 6 m long and equipped with two pressure
sensors. A recording system coupled to a computer allows
the recovery of the pressure sensors signal and the control of
the pressure evolution across the test section in real time.
The test procedure consists essentially to follows in real
time the variation of the pressure drop of the heavy oil flow
with and without lubrication, in order to observe the
behaviour and the efficiency of the lubrication process along
Crude oil supply tank and the separation system of crude oil from water of lubrication
Fluid of lubrication supply tank
Fluid of lubrication injector
6 EXPERIMENTAL RESULTS AND DISCUSSION
Before the presentation of the pipeline lubrication results,
rheological characterisation has been carried out on the
oil using the flow loop. The results have been compared with
the measurements from a rotary viscometer (Rheometrics
RFS II) in order to validate the metrology of the installation
used to perform this study.
6.1 Rheological Study and Validation of the Metrology Used
Evolution of the pressure drop in the loop with the increase
of the flow rate is presented on the Figure 7. A linear
evolution is observed which is characteristic of laminar flow
Wall shear stress and shear rate variation can be deduced,
according to the following relations:
D ∆ P
τ = ⋅ : shear stress (Pa);
γ˙ = = : shear rate (s–1);
∆ P/L linear pressure loss (Pa/m);
D diameter of the conduct (m);
Q total volumetric flow rate (m3/s);
V flow velocity (m/s).
The observed linear evolution of the shear stress versus
the shear rate corresponds to a Newtonian behaviour (τ = µ ·γ· )
with a viscosity of the heavy oil µ = 4.74 Pa·s at a
temperature of 19.7°C.
In order to validate the metrology of the installation used
to perform the lubrication tests, the rheological results
obtained from the flow loop, were compared to those
obtained from the rotary viscometer ‘’Rheometrics RFS II’’.
The result of this comparison is presented on Figure 8, and
showed a perfect concordance between the results obtained
from the flow loop experiments and those obtained from the
rotary viscometer, which validate the metrology of the
6.2 Lubrication Process
6.2.1 Example of Data Acquisition
Variation of the pressure drop of the heavy oil with and
without lubrication is measured in order to analyse the
behaviour and the efficiency of the lubrication process.
An example of data acquisition is given on Figure 9 and
shows the influence of the injection of water as lubricant,
during the transport of the heavy oil. The applied conditions
used for this tests are (Fig. 9a) heavy oil flow rate = 293 l/ h
and a flow rates ratio between water and oil of 6%.
The first plateau corresponds to the pressure drop due to
the heavy oil flow before injection of the water in the pipe
(0.39 bar/m). 250 s after the beginning of the test, the
lubrication fluid is injected. A fall of the pressure loss in the
measurement section is observed and pressure drop stabilises
around 0.01 bar/m after 350 s. The flow is then lubricated
along of the entire pipe. The superficial velocity for the oil
Comparison between the results obtained from the flow loop and
those given by the rotary viscometer ‘’Rheometrics RFS II’’.
phase in these conditions is 0.16 m/s. If we consider a linear
velocity profile for the water in the lubricated film, the mean
velocity is then roughly the half of the oil superficial
velocity. The lead to around 140 s to fill the entire length of
the pipe with the water film. This order of magnitude is in
very good agreement with the time that it takes to reduce the
friction. The injection of the lubrication fluid is stopped at
t = 950 s, and with roughly the same time scale, the pressure
drop in the pipe increase rapidly to reach its initial value of
Another example is given in Figure 9b where the variation
of the pumping pressure versus the time are represented for
the following conditions:
– heavy oil flow rate = 340 l/h;
– ratio between the flows rate of water of lubrication and of
heavy oil used = 0.06;
– temperature = 20.1°C.
6.2.2 Effect of the Quantity of the Lubrication Fluid and the Flow Rate of Oil
Lubrication fluid optimisation is achieved by studying the
effect of the quantity of the water on the lubrication process
efficiency. In these conditions, the parameter is the ratio
(Qw/Qo) between the flow rates of water to oil.
Figure 10 gives the variation of the pressure drop
reduction versus water to oil flow rates ratio. These results
show that an increase of the pressure drop reduction is
observed when we increase the water flow rate. This increase
becomes negligible from a value of flow rate ratio of about
6%. Therefore, we can conclude that an optimal lubrication
rate associate to a minimal use of the lubrication fluid could
be obtained for a flow rate ratio of 6%.
6.2.3 Pressure Drop Variation with Flow Rate with and without Lubrication
The pressure drop in the pipe is measured for different oil
flow rate with and without lubrication. Results are presented
in Figure 11. Without lubrication, we showed that a linear
increase of the pressure drop is observed in very good
agreement with the Poiseuille law and an oil viscosity
of 4.74 Pa·s.
For lubricated flow, the flow rates ratio is fixed at 6% and
we increase the oil flow rate (and consequently the water
flow rate). Decrease of the pressure drop with flow rate is
observed as shown in Figure 11.
This result is remarkable because pressure drop is always
increasing with flow rate in usual conditions. The increase
rate can be very different for diverse conditions (laminar or
turbulent, Newtonian, shear thinning or shear thickening,
etc.), but a decrease is rarely observed.
This property is in fact a particular feature of CAF when
difference of density between water and oil is present.
Pressure Drops of the Lubricated Flow
If we assume a “perfect” core annular flow, well centred, as
represented in Figure 2, pressure drop can be calculated from
the following relation:
π R 4
8 µ w
pressure drop of the perfect core annular flow (Pa/m);
total flow rate (m3/s);
pipe radius (m);
Example of data acquisition of the heavy oil lubrication test.
Qo = 340 l/h
Qo = 293 l/h
Qo = 230 l/h
5 6 7 8
Flow rates ratio (Qw/Qo)(%)
Effect of the ratio between the flow rates of the water and the
heavy oil on the efficiency of the lubrication process.
We notice that the pressure drop reduction is calculated
according to the following relations:
Pressure drop reduction (%) =
∆ P : pressure drop of the lubricated flow (bar/m).
Pressure drop of crude oil in flow without lubrication
Pressure drop of crude oil in lubricated
flow using water with a ratio of 6%
200 220 240 260 280 300 320 340 360
Heavy crude oil flow rate (l/h)
As we already mentioned, it’s mainly dependent on water
viscosity and water layer at the wall.
When lubricated and core centred in the pipe, the pressure
drop can be calculated with Equation (2). Maximum flow
Reynolds number for water is around 1500, so we assume
that the flow is laminar and that Equation (2) can be applied.
Comparisons between calculated pressure drop, and
experimental measurements show a large discrepancy as one
can see on Figure 12. The pressure drop in the experimental
tube is much larger than expected.
As we already pointed out, increase of pressure drop
with oil flow rate is attended from Equation (2). But
experimentally, the pressure drop is decreasing with oil flow
rate if we fixed the ratio of water to oil flow rate to a fixed
value (6% in Figure 11).
These surprising results can be explained if we take into
account the density difference between the phases. Indeed,
in the case of heavy oil transport, density difference tends to
lift the core in the upper part of the pipe (oil density is
800 kg/m3). A schematic of the radial position of the core is
given in Figure 13. Lubricating forces tend to push the core
in the centre. A competition between the two effects gives a
position of the core varying from completely centred, when
lubricated force are large (for large velocity) to completely
excentred when the core is in the upper part of the pipe
touching the wall, when velocity is small. All positions in
between are possible.
A simple phenomenological model is proposed to take
into account this effect and we will show that in doing so, the
experimental observation can be explained.
For liquids flowing in tube, pressure drop is balanced by
wall shear stress in steady state conditions. In the case of
partially centred annular flow, we divide the pressure drop in
two terms: a component due to the water, and a component
due to the oil.
In introducing S, the mean contact perimeter between
the core of the flow (oil) and the wall of the pipe, we
approximate the pressure drop as the simple expression:
∆ LP = 8π µ w R 4 Q– Rs4 ξ( – 1) + 8π µ o RQ4 ξ( ) withξ = SS0
S0 = 2π·R is the perimeter of the pipe, and S is the contact
perimeter between the pipe wall and the core of the flow
(heavy oil). The model depends only on the ratio ξ.
ξ is depending on the density difference, but also on the
amount of water injected. For a fixed oil velocity, ξ is
decreasing with the amount of water. And for a fixed amout
of water, ξ is decreasing with the oil velociy, due to the
increase of lubricated effect.
A simple phenomenological function can be choosen to
have this behaviour:
where K is depending upon density difference, but is fixed in
Qualitative estimation of the pressure drop with Equations
(4) and (5) were done. Pressure drop reductions for different
oil flow rate and pressure drop with oil flow rate are shown
in Figure 14.
The tendencies are in very good agreement with the
experimental observation. For a fixed ratio of water to oil
flow rate, the pressure drop is decreasing with oil flow rate
because the lubricated forces is increasing with velocity and
push the core in the centreline. In doing so, water can
lubricate all the perimeter of the core and pressure drop is
lower. This effect last until the core is well centred in the
Quantitative calculations were not satisfying with the
simple solution proposed in Equation (5) as we can check in
In order to evaluation in more detail the ξ dependency
with velocity and amount of water, we try to estimate it from
6.2.4 Perimeter Measurement
Starting from the experimental measurements, we can
calculated the values of ξ that give the presures drop. S is put
into Equation (4) to fit the experimental measurements of the
pressure drop. In doing so, we get the results reproduced in
the Table 1 below.
These results are represented on Figure 16. Evolution of ξ
with the two main parameters of our system is shown
(amount of water and oil velocity).
From theses results one can observe a decreasing of the
ratio ξ, with an increasing of the fluid forming the core (oil)
flow rate and/or the amount of the fluid lubricant (water). As
Model calculation of pressure reduction function of amount of water and pressure drop with oil flow rate for a fixed quantity of water.
Flow rates ratio (%) 10 12 14
we discussed before, at large oil velocity, the core is almost
centred and ξ is around 0.
Starting from the parameter envelope calculated and
presented in Figure 16, we can determine a correlation
function that can reproduce the evolution of ξ = S/S0. We can
then estimate the pressure drop for heavy oil in large-scale
system taking into account the difference of density of the two
phases and the fact that the core is not centred in the tube.
S/S0 value calculated to adjust experimental pressure drop results
Heavy oil transportation through pipeline is challenging
because of its huge viscosity. In this paper we investigated
experimentally a technique based on lubricated flow. The
heavy oil is injected at the centre of the pipeline and a small
water film is co-injected around it. When water start to be
injected, the flow regime converge rapidly to Core Annular
Flow and keep stable in time. The flow rate of water is a
fixed ratio of the oil flow rate. Pressure measurement shows
a reduction of more that 90% of the pressure drop with
lubrication. Variation of the flow rate ratio is done and we
observed than for more than 6% of water flow rate, the
pressure reduction reach a plateau depending on the oil flow
rate. Reduction is larger for higher flow rate. Variation of the
pressure drop with increase of total flow rate is also
performed. A very curious observation is made: a decrease of
the pressure drop with the increase of flow rate.
To estimate the reduction in pressure drop, the calculation
of the pressure drop in the case of perfect core annular flow is
done. A large discrepancy is found between calculation and
experiment. A simple phenomenological model is then
proposed to take into account density difference between oil
and water. The buoyancy force tends to lift the core at the top
of the tube and the lubrication process tends to centre it.
Competition between the two mechanisms lead to an
eccentric position of the core and a small part of the oil core is at
the wall leading to a contribution of the pressure drop. An
empirical estimation of the oil wet perimeter allows the
prediction of the decrease of the pressure drop with flow rate and
the dependency of the pressure reduction with oil flow rate.
The huge reduction of friction in pipe gives a large
industrial interest for this lubricated technique to transport
heavy oil. Steady state flow has been demonstrated in this
study and the pressure reduction can reach 95% for some
flow parameters. Stop and restart have to be studied to
guaranty the complete feasibility for an industrial scale.
Oil Industry. Proceedings of the 15th World Petroleum
01 Isaacs, J.D. and Speed , J.B. ( 1904 ) Method of Piping Fluids . US Patent N° 759374 .
02 Oliemans, R.V.A and Ooms , G. ( 1986 ) Core - Annular Flow of Oil and Water through a Pipeline, in: Multiphase Science and Technology. Hewitt G.F. , Delhaye , J.M. and Zuber , N., 2 , Washington Hemisphere .
03 Joseph, D.D, Chen , K.P. and Renardy , Y.Y. ( 1997 ) Core Annular Flows . Ann. Rev. Fluid Mech., 29 , 65 .
04 Govier, G.W. and Aziz , K. ( 1972 ) The Flow of Complex Mixture in Pipes , Von Nostrand Reinhold Company.
05 Joseph, D.D. and Renardy , Y.Y. ( 1993 ) Fundamentals of Two-Fluids Dynamics , New York Springer-Verlag.
06 Bai, R. , Chen , K. and Joseph , D.D. ( 1992 ) Lubricated Pipelining: Stability of Core-Annular Flow. Part 5. Experiments and Comparison with Theory . Journal of Fluid Mech. , 240 , 97 - 132 .
07 Ooms, G. , Segal , A. , Van der Wees , A.J. , Meerhof , R. and Oliemans , R.V.A. ( 1984 ) A Theoretical Model for CoreAnnular Flow of a Very Viscous Oil Core and a Water Annulus through a Horizontal Pipe . Int. Journal of Multiphase Flow , 10 , 1 , 41 .
08 Prezioki, L. , Chen K. and Joseph , D.D. ( 1989 ) Lubricated Pipelining: Stability of Core-Annular Flow . Journal of Fluid Mech. , 201 , 323 - 356 .
09 Guevara, E. , Gonzales , S.A. and Nunez , G. ( 1998 ) Highly Viscous Oil Transportation Methods in the Venezuelan
10 Arney, M.S. , Bai , R. , Guevara , E. , Joseph , D.D. and Lui , K. ( 1993 ) Friction Factor and Holdup Studies for Lubricated Pipelining - Experiments and Correlations . Int. J. Multiphase Flow , 19 , 6 , 1061 - 1076 .
11 Huang, A. , Christodoulou , C. and Joseph , D.D. ( 1994 ) Friction Factor and Holdup Studies for Lubricated Pipelining - Laminar and k-ε Modes of Eccentric Core Flow . Int. J. Multiphase Flow , 20 , 3 , 481 - 491 .
12 Guevara, E. , Zubillaga , V. , Zagustin , K. and Zamora , G. ( 1991 ) Core Annular Flow: Preliminary Operational Experience and Measurements in a 152 mm*55 km Pipeline . Proc. 5th Unitar Heavy Crude and Tar Sands International Conference , 5 , Caracas, Venezuela.
13 Joseph, D.D. , Bai , R. , Mata , C. , Sury , K. and Grant , C. ( 1999 ) Self-Lubricated Transport of Bitumen Froth . Journal of Fluid Mech. , 386 , 127 .
14 Arney, M.S. , Ribeiro , G.S. , Guevara , E. , Bai , R. and Joseph , D.D. ( 1996 ) Cement-Lined for Water Lubricated Transport of Heavy Oil . Int. Journal of Multiphase Flow , 22 , 2 , 207 .
Final manuscript received in September 2004