OTC 17081
Studies on the Mechanisms of Wax Deposition in Pipelines
A.T. Leiroz and L.F.A. Azevedo, Dept. of Mechanical Engineering, PUC-Rio
Copyright 2005, Offshore Technology Conference
This paper was prepared for presentation at the 2005 Offshore Technology Conference held in
Houston, TX, U.S.A., 25 May 2005.
This paper was selected for presentation by an OTC Program Committee following review of
information contained in a proposal submitted by the author(s). Contents of the paper, as
presented, have not been reviewed by the Offshore Technology Conference and are subject to
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Offshore Technology Conference, its officers, or members. Papers presented at
OTC are subject to publication review by Sponsor Society Committees of the Offshore
Technology Conference. Electronic reproduction, distribution, or storage of any part of this
paper for commercial purposes without the written consent of the Offshore Technology
Conference is prohibited. Permission to reproduce in print is restricted to a proposal of not
more than 300 words; illustrations may not be copied. The proposal must contain conspicuous
acknowledgment of where and by whom the paper was presented. Write Librarian, OTC, P.O.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.
Abstract
Deposition of high molecular weight paraffins on the inner
wall of subsea production and transportation pipelines
continues to be a critical operational problem faced by the
petroleum industry. The accurate prediction of wax deposition
rates and deposited wax spatial distribution would be
invaluable information for the design of subsea lines. A
critical review of the literature conducted lead to the
conclusion that there is not enough experimental evidence to
determine which are the relevant mechanisms responsible for
wax deposition. Based on the conclusions of the literature
search, a research program was initiated with the aim of
identifying the relative importance of the suggested wax
deposition mechanisms namely, molecular diffusion,
Brownian diffusion, shear dispersion and gravity settling. To
this end, simple basic experiments were designed and
complemented by numerical simulations. In these
experiments, visualization of the deposition process was
sought for a stagnant solution of paraffin and solvent within a
cavity submitted to a transverse temperature gradient. Studies
were also conducted for deposition under laminar channel
flow, also submitted to a transverse temperature gradient. The
visualizations studies for the deposition in the stagnant layer
revealed the formation of a mushy region growing from the
cooled wall formed by a network of crystals. Images with
higher magnifications revealed motions of crystals not formed
at wall. Detailed data of temporal and spatial distributions of
the wax deposited under laminar flow conditions were
obtained. Numerical simulations of the experiments conducted
were developed employing molecular diffusion as the only
deposition mechanism, as it is done in the vast majority of the
deposition models available in the literature. The model under
predicted the evolution of the deposited wax, indicating that
mechanisms other than molecular diffusion may be present.
Introduction
Wax deposition in production and transportation pipelines
continues to be a relevant problem for the industry,
particularly in offshore operations. The crude oil flows out of
the reservoir at, typically, 60 oC into the production pipelines.
These lines carry the oil to the platforms and from the
platforms to shore. At large water depths, the ocean
temperature at the bottom is of the order of 5 oC. The
solubility of wax in the oil is a decreasing function of
temperature. As the oil flows, it loses heat to the surrounding
water. If the crude oil temperature falls below the Wax
Appearance Temperature (WAT), the wax may precipitate and
deposit along the inner walls of the pipeline. The accumulation
of the deposited material may lead to increased pumping
power, decreased flow rate or even to the total blockage of the
line with loss of production and capital investment.
A significant research effort has been devoted to the
understanding and modeling of the wax deposition problem
[1-4]. This is a complex problem that involves several
disciplines such as thermodynamics, heat transfer, mass
transfer, crystal growth and fluid dynamics. An accurate
prediction of the temporal and spatial distributions of the
deposited wax along the pipeline would be invaluable
information that would help in the design stages of the field, as
well as in the scheduling of interventions in the pipeline, in
order to assure the flow of oil at the desired rates.
A detailed critical review of the literature was prepared as
part of a research project on wax deposition [3]. One of the
key works analyzed in this review was that of Burger et al. [1]
where the possible mechanisms responsible for wax deposition
were identified. These mechanisms are molecular diffusion,
Brownian diffusion, shear dispersion, and gravitational
settling. Molecular diffusion has been widely accepted as the
dominant deposition mechanism, and has been included in the
vast majority of the models presented in the open literature
[e.g., 6,7]. However, in the review article mentioned above
Azevedo et al. concluded that there was no experimental
evidence that confirmed molecular diffusion as the dominant
deposition mechanism. Indeed, in a recent paper [7] presents a
mathematical model for wax deposition that considers both
molecular diffusion and shear dispersion as the relevant
mechanisms.
An observation of the deposition models available in the
literature shows that they all make use of adjusting constants
to fit the model predictions to field or laboratory data. In our
view this practice, totally justified for yielding good
2 OTC 17081
predications for particular lines, difficult the task of assessing
the relative importance of each deposition mechanism.
The present paper presents results of an ongoing research
effort aimed at identifying the relative importance of the
deposition mechanisms. To this end, small-scale, laboratory
tests were conducted employing simple oil-wax mixtures with
known properties. The test sections constructed were simple,
having well-defined boundary and initial conditions. The
experimental tests were complemented by numerical
simulations that allowed the testing of different deposition
models. In the simulations, the measured properties of the test
fluids were employed. No constants were adjusted to fit the
predicted results to the measured data. It is believed that the
comparison between the experimental and predicted results for
these simple configurations would allow a better assessment of
the relative importance of the deposition mechanisms. In the
results presented here, only the molecular diffusion-based
mechanisms will be studied.
As will be described next, two types of tests were
conducted. Firstly, the depositions tests were conducted for a
stagnant fluid layer within a sealed cavity submitted to a
temperature gradient. After that, tests were conducted for wax
deposition under laminar flow within a rectangular channel. In
both cases, the experiments were designed so as to allow the
visualization of the deposition process. The test sections
employed are now described in details.
Experiments with deposition in stagnant fluid
The test section was designed to allow the study of wax
deposition in a stagnant solution. Figure 1(a) presents a
schematic view of the experimental apparatus. The cavity was
formed by two 2-mm-thick vertical copper walls and two
horizontal glass walls. The copper walls were fins attached to
copper hollow blocks. Each copper block was connected to a
temperature controlled water-circulating bath. The lower
horizontal glass wall was constructed by a sandwich of two
glass layers in order to decrease the internal width of the
cavity to a dimension of 0.5 mm. This small dimension was
necessary to eliminate any natural-convection-induced motion
within the cavity. Two vertical, lateral walls and rubber
gaskets sealed the cavity. In one of the lateral walls small
holes were made to allow the filling of the cavity with the
working fluid and the removal of the trapped air. A
hypodermic needle and syringe were used in the filling
operation. A fine-gauge thermocouple was installed inside
each copper fin, near the inner cavity surface. These
thermocouples were used to monitor the temperature history of
the two cavity surfaces. The final internal dimensions of the
cavity were 10 x 30 x 0.5 mm (width x length x thickness). A
view of the assembled cavity is shown in Figure 1(b).
A digital camera with resolution of 640 x 480 pixels and
equipped with an 18108 mm zoom lens was mounted
vertically with its axis orthogonal to the cavity horizontal glass
wall. White back illumination was employed. The camera was
connected to a video recorder to capture the evolution of the
wax deposition front. The videos recorded were digitized and
an image processing software was used to measure the
thickness of the deposited layer as a function of time. In order
to control the temperature of the external environment, the
cavity was housed in a Styrofoam box that only allowed
access to the camera lens and back illumination. A liquid-in-
glass mercury thermometer monitored the internal temperature
within the box.
The preparation for a data run was initiated by filling the
cavity with the paraffin solution to be tested. For the tests
presented in the present paper a 10% in volume solution of
refined paraffin with carbon numbers between C21 and C38
dissolved in n-paraffin (C13 to C16) was used. The WAT for
the utilized solution was measured to be of the order of 27 oC.
Prior to the filling operation, the water circulating units
were set to a temperature above the WAT of the testing
solution, in order to avoid unwanted depositions in the filling
tubes and inside the cavity. The cavity was filled with the
testing solution and positioned horizontally on a support. After
a steady state condition was verified by the thermocouple
readings, the data run was initiated by the imposition of a
stepwise change on the temperature of one of the copper walls.
This was achieved by a fast connection of one of the copper
walls to a circulating water bath that had been previously
equilibrated at the lower temperature desired for the test being
conducted. This lower temperature was always lower than the
WAT for the solution tested. The opposite copper wall was
kept at the elevated temperature. The evolution of the
deposition front was continuously recorded by the video
system until a steady state position of the front was reached.
After the test was completed, the temperature of both copper
walls were elevated above the WAT in order to dissolved all
the deposited wax, thereby preparing the test section for a new
data run.
As the tests were conducted, it was found necessary to
obtain images of the deposition front using higher
magnifications. To this end, the test section described was
mounted on a microscope table. The small working distances
of regular microscope objective lenses did not allow that
magnifications higher than 50x were obtained. Objectives with
higher magnifications have smaller working distances and
interfered with the higher parts of the test section.
In order to allow the observation of the deposition
phenomena employing higher magnifications, a new test
section was designed and constructed. This test section of
smaller dimensions is presented in Figure 2. The basic design
conception of the new test section is the same as the previous
one: a stagnant fluid sealed within a cavity having two end
walls kept at different temperatures. The cavity employed was
made from a commercially available glass tube with parallel
walls. The internal thickness of the tube was 0.7 mm, and its
width was 10 mm. The length of the tube was cut to 40 mm.
The end walls of the glass tube were closed by copper fins that
penetrated 5 mm in the tube. These copper fins were soldered
to hollow copper blocks connected to the water thermostatic
baths and circulating units. Fine gauge thermocouples were
installed in the copper fins to monitor the temperature of the
walls.
The cavity filled with the test solution was positioned on
the motorized microscope table. The data run was initiated by
heating both copper walls to a temperature level above the
WAT for the particular solution being tested. After the steady
state condition was reached, the temperature of one of the
walls was lowered, while the other was kept at the elevated
OTC 17081
level. A cooling ramp was imposed on the cooled wall. Since
the field of view of microscope lenses is rather small, the
imposition of a step change on the wall temperature, as it was
described previously for the larger cavity experiments, would
make the deposition front too fast and nearly impossible to be
observed in the small field of view. The use of the cooling
ramp allowed the control of the deposition front speed and its
observation with higher magnifications. In some cases, the
deposition front was observed for longer periods of time by
moving the test section under the microscope objective at the
approximate front speed. This was achieved by using the
computer-controlled, motorized microscope table.
Experiments with deposition under laminar flow
The experiments on wax deposition under laminar channel
flow were conducted employing the test section that will now
be described. Figures 3 and 4 show schematic views of the test
section.
The heart of the test section was a rectangular channel with
internal dimensions of 3x 10 x 300 mm (width x height x
length). The vertical walls of the channel were made of 2-mm-
thick glass plates. The upper and lower walls were made of
copper and were soldered to hollow copper blocks. Water
from a temperature controlling and water circulating unit was
pumped through each of the copper blocks. Fine gauge
thermocouples were installed in the copper walls to monitor its
temperature. Each end of the rectangular channel was
connected to a stainless steel tank. The larger upstream tank
contained the solution to be tested. In order to control the
solution temperature, the upstream tank was equipped with a
coil-tube heat exchanger connected to a water circulating and
thermostatic unit. The smaller downstream tank was equipped
with an electric heater that was used to avoid the deposition of
the wax in the tank. A progressive cavity pump was used to
return the solution from the downstream tank to the upstream
tank, thereby allowing for a continuous operation of the test
section.
A 640 x 480 digital camera equipped with a long-working
distance microscope objective was used to register de
evolution of the wax deposition process inside the channel.
The camera was mounted on a X-Y coordinate table, in such a
way that the camera was positioned with its axis orthogonally
to the glass wall of the channel. The camera could be moved
along the axial and cross-stream directions of the channel. In
the cross-stream direction a precision screw allowed the
motion of the camera within 1/100 mm. That was necessary to
adjust the focus of the image. Back illumination through the
opposite glass wall of the channel was employed. A 20x
magnification microscope objective was used.
A data run was initiated by circulating the oil-wax solution
to be tested through the channel and back to the tank in a
closed loop. The temperature of the solution, as well as the
temperature of water circulating inside the copper blocks, was
above the WAT for the solution. The mass flow rate desired
for the particular experiment was set in the pump. After a
steady state condition was achieved, the temperature of the
water in the copper blocks was lowered to a value below the
WAT. The deposition of the wax along the lower and upper
3
copper walls was registered by a tape recorder connected to
the digital camera.
At any fixed axial position of the camera, the temporal
evolution of the deposited wax could be obtained by the
analysis of the recorded images. However, the field of view of
the lenses employed only allowed the visualization of an area
of approximately 15 x 15 mm2. This area was sufficient for
visualizing the whole 10-mm height of the channel. The
channel length of 300 mm could not be observed in such a
small field of view.
A special experimental procedure was devised to allow the
visualization of the deposition along the whole length of the
channel. Firstly, the camera was positioned at the entrance of
the channel, imaging the first 15 mm of its length. The
transient evolution of the deposited layer at this axial position
was registered by the camera up to the attainment of a steady
state condition for the deposited thickness. Wax deposit
growth at both upper and lower walls could be registered
simultaneously in the same image fields. After attainment of
the steady state condition for the particular mass flow rate
being studied, the temperature of the copper walls was raised
to a value above the WAT. With this operation, all the
deposited wax was removed. The coordinate table was then
moved to a new axial station, adjacent to the previous one.
After attainment of steady state conditions, the temperature of
the copper walls were lowered to the same values used in the
experiments for the previous position of the camera. The
transient evolution of the deposit wax layer was registered
until a steady condition was reached. This procedure of
forming and removing the wax deposit while moving the
camera was repeated until the whole channel length was
visited.
In order to estimate the accuracy of the measuring
procedure just described, a series of test experiments was
conducted prior to the data runs. In this test experiments, the
camera was kept fixed in a particular axial position and
registered the time evolution of the deposit. After steady state
the deposit was removed by increasing the wall temperature.
The experiment was then repeated without, however, moving
the camera. The results of several replications were compared.
In all cases the deviations were within 5%, what was
considered satisfactory.
Numerical model for deposition in the stagnant layer
As already mentioned in the introduction, the numerical
simulation of the wax deposition is an important part of the
present research. The role of the numerical solution developed
in the present work was to help identify the relative
importance of the wax deposition mechanisms proposed in the
literature. The idea was to develop a simple model that
incorporated one deposition mechanism at a time and compare
the predictions obtained with the simple experiments
conducted. In the present paper, only a model for deposition
by molecular diffusion was implemented, since this is the most
widely accepted mechanism.
The simulation of the deposition in a stagnant layer
assumed the following hypothesis:
4 OTC 17081
(i) One dimensional heat conduction and mass
diffusion
(ii) Constant properties. Thermal conductivity,
specific heat and density are the same for liquid
and solid phases
(iii) Saturated liquid at the solid/fluid interface
With the assumptions listed above, the energy equations for
the liquid and solid phases can be written as,
For the solid region:
2 Ts 1 Ts
= (1)
y s2 t
For the liquid region:
2 Tl 1 Tl
= (2)
y l2 t
At the solid/liquid interface, an energy balance yields:
Ts T dS (t )
k k l = w L (3)
y s y l dt
In the above equations Ts and Tl are, respectively, the
temperatures in the solid and liquid phases, ys and yl are
spatial coordinates of the solid and liquid phases, S(t) is the
position of the interface, t is the time, is thermal diffusivity,
and is the solid deposit porosity defined as the ratio of the
mass of liquid to the total mass of solid and liquid. The density
of the solid paraffin is w, while its latent heat of fusion is L.
The solute (paraffin) mass fraction distribution is governed
by,
2 ww 1 ww
= (4)
y l2 D t
At the interface, a mass balance produces a relation for the
velocity of the interface,
dS (t ) D T ww
= transverse concentration gradient is evaluated by the
dt (1 ) w y l transverse temperature gradient and a solubility parameter.
(10)
In the equations above, ww is the mass fraction of paraffin
in the liquid phase, D is the molecular diffusion coefficient,
and T is the density of the solution. By assuming
thermodynamic equilibrium at the interface the mass fraction
at the interface is determined as a function of the interface
temperature by means of a solubility curve determined
experimentally.
As initial condition, the solution is assumed to be at a
uniform elevated temperature, Tq. For t>0, a colder
temperature, Tf, is imposed at the cold wall, y = 0, while the
hotter wall at y = H, is maintained at Tq. The initial mass
fraction wwo is imposed as equal to 1 in the solid phase
(0<y<S(t)) and equal to 0,1 in the liquid phase (S(t)<y<1). The
hotter wall at y = H was considered impermeable, so
ww / y = 0 . At the solid/liquid interface, y = S(t), the solid
and liquid temperatures were equal, i.e., Tl(y,t) =Ts(y,t), and
the mass fraction was the equilibrium mass fraction for the
interface temperature, w = wint.
The governing equations were written in dimensionless
form using the following dimensionless variables: y* = y/H, S*
= S(t)/H, t* = t/H2, and = (T Tf)/(Tq Tf). The
dimensionless parameters governing the problem are the
Stefan number, Ste = CP(Tq Tf)/L, the Lewis number Le =
/D, the ratio of the solid paraffin to the solution density, =
w/T, and the porosity, .
The set of equations described, together with initial and
boundary conditions, were solved by a finite difference
method. Since the paraffin deposition is a moving boundary
problem, a coordinate transformation was employed to
transform the variable physical domain into a fixed
computational domain. Details of the transformation employed
and of the implementation of the numerical method can be
found in [8].
Numerical model for deposition in laminar channel
flow
The numerical model developed for simulating wax
deposition under laminar channel flow conditions followed the
models normally described in the literature (e.g., [6]). In these
models the flow is treated as one-dimensional, with the axial
velocity varying with the axial coordinate due to the decrease
in channel cross sectional area caused by the wax
accumulation. A local varying friction factor for fully
developed flow was employed. The energy equation was
solved to yield the axial variation of the bulk fluid
temperature. Heat losses through the transverse channel
direction were computed taking into account the convective
thermal resistance within the channel and the conductive
resistance in the deposited layer and channel metallic wall.
Heat losses through the glass lateral walls were also
considered.
The growth of the deposited layer was accounted for by a
molecular diffusion mechanism, as suggested by Burger et al.
[1]. In this model, the diffusion flux of wax toward the cold
wall was estimated by Ficks law of diffusion, where the
Results and discussion
The results obtained in the present study will now be
presented. Figures 5(a) to 5(d) is a sequence of images
captured of the evolution of the deposition front for a typical
experiment. In the images, the top wall is the hot wall kept at
32 oC, while the bottom wall is the cold wall kept at 12 oC.
The images correspond to, respectively, 0, 15, 48 and 125
seconds after the initiation of the cooling of the lower wall.
The external ambient temperature was 25oC. The evolution of
the deposition front can be clearly observed in the images.
Images like the ones presented in Figure 5 were processed to
OTC 17081
yield quantitative information of the time variation of the
deposition front.
The level of magnification used in the images shown in
Figure 5 did not allow the observation of the details of the
formation of the wax crystals close to the cooled wall. Images
with higher levels of magnification, namely, 50, 200 and
1000x, were obtained.
Figure 6 shows images with 200x magnification. In the
figure the cold wall is at the left and the hot wall is out of the
field of view to the right. The formation of nearly hexagonal
crystals both at the wall and in the vicinity of the wall are
observed. The observation of the live tapes that generated the
pictures indicate that the crystals formed in the vicinity of the
wall are freely moving and turning, up to the time when they
are trapped in the crystal network formed. No clear, consistent
crystal motion toward the wall was observed.
At the even larger magnification of 1000x the growth and
motion of one single crystal toward the cooled wall can be
observed. Figure 7 display a sequence of such images where
the left wall is the cold wall presenting some crystals already
attached to it. A single crystal is observed growing and
moving toward the wall.
Figure 8 shows the quantitative results of the deposition
thickness of the wax layer for the stagnant fluid in the cavity.
In the figure the abscissa is the time given in seconds, while
the ordinate represents the thickness in millimeters. Each
curve in the figure represents a different value of the external
ambient temperature. An observation of the figure shows that
the deposited layer grows at a faster rate at the beginning of
the experiment, leveling off to reach an asymptotic value for
longer times. During the initial transient, a data spread can be
noticed, indicating that the external temperature is influencing
the rate of deposit growth. The lower the external temperature,
the higher the heat losses from the cavity, leading to an
increase in the deposit growth rate. Nevertheless, it is seen that
the asymptotic value attained is not influenced by the heat
losses, as can be verified in the figure.
Also shown in Figure 8 are curves that represent the
predictions of the evolution of the deposited layer obtained
with the numerical model developed as part of this study.
Three curves are presented for different values of the Lewis
number, namely, 2.96, 29.6, and 296. The value of Lewis
number equal 296 was obtained employing the actual
properties of the solution utilized in the tests. The calculations
with the other two values of Lewis were conducted to assess
the sensitivity of the solution to this dimensionless parameter.
In all cases the Stefan number was 0.19.
A general observation the numerical results of Figure 8
shows that the numerical solutions under-predict the
experimental data by about 30 %. The influence of the Lewis
number does not seem to be significant. The results presented
in Figure 8 seem to be the first comparison available in the
open literature between measured and predicted paraffin
deposition. As already commented in the Introduction, all the
models available in the literature make use of adjustments of
solution properties mainly the coefficient of diffusion in
order to better reproduce the experimental data. In the present
research, the use of actual properties allows the assessment of
the use of molecular diffusion as the only mechanism
responsible for wax deposition. The fact the diffusion-based
5
model under-predicts the deposition thickness is an indication
that other deposition mechanisms may be present. Other
factors not related to the deposition mechanisms may be
affecting the agreement between experiments and predictions.
Heat losses from the glass and lateral walls of the cavity have
not been considered in the one-dimensional model developed.
At the time this article was being prepared, heat loss
mechanisms were being included in the model developed.
We now focus our attention in the results obtained for wax
deposition under laminar channel flow. Figure 9 display
typical results obtained. In the case of the figure, The
Reynolds number for the flow is equal to 580. The inlet
temperature of the solution was 40oC, while the cold walls
were maintained at 15oC. The WAT for the solution used was
36oC.
The results presented in Figure 9 are for steady state
conditions of the deposited layer. The black squares in the
figure are the deposition thickness measured from the images
for several axial positions of the camera, as described
previously in the experiments session. The solid lines in the
figure are just fitted lines to help interpret the trends displayed
by the data. As can be seen in the figure, the deposition on
both the upper and lower cooled walls could be measured in
the same experiment. The excellent symmetry of the data is an
indication of the symmetry on the velocity and temperature
profiles within the channel, which attests for the quality of the
test section constructed. Another important conclusion that can
be drawn from the excellent symmetry presented by the
deposits on both walls is that gravity settling is not a relevant
deposition mechanism. Burger et al., [1], have already
presented this conclusion. However, they reached it via an
indirect way, by comparing deposition results in vertical and
horizontal pipes. The results presented here are the first direct
measurement and visualization of this conclusion.
The experimental methodology used in the present work
allowed the determination of the spatial and temporal
evolution of the wax deposits. Figures 10 and 11 show typical
results obtained. In both figures the thickness of the deposited
layer given in millimeters is presented as a function of the
axial coordinate of the channel, given also in millimeters.
Each curve represent a different time counted from the
initiation of the cooling of the walls. The time intervals range
from 1 minute to 4 hours. At four hours, no more deposit
growth was verified, indicating the attainment of a steady
condition. In both cases presented in the figures, the fluid inlet
and wall temperatures were 40 and 15oC respectively. The
WAT was equal to 36oC. In the case of Figure 10 the
Reynolds number was 301, while in Figure 11 it was 729.
The data presented in Figures 10 and 11 are, seemingly, the
first published results that display the spatial and temporal
variation of the wax deposited layer in channel flow.
An observation of Figure 10 reveals the rapid growth of the
deposit layer. Indeed, the first 10 minutes of deposit
accumulation are responsible for nearly 50% of the final,
steady state thickness. A comparison of the results from Figure
10 and 11 show that, as expected, the deposited layer is
comparatively thinner for the higher flow rates represented by
the higher value of the Reynolds number.
Figure 12 presents a comparison of the deposition results
obtained from the diffusion-based numerical model developed
6 OTC 17081
with the experimental data. The results are for Reynolds
number equal to 301. The actual properties available for the
test fluid employed in the experiments were used in the
calculations. In Figure 12 the symbols represent the
experimental deposition thickness measured at the times
indicated in the legend. The upper most data in the figure
represent the steady state regime. The solid lines are the
predictions from the model for the same times as those
measured in the experiments. These lines are color-coded to
the experimental data points.
A general observation of Figure 12 shows the poor quality
of the model predictions. Indeed, except for the steady state
case, the model was not able to even predict the correct trend
of the spatial distribution of the deposited layer. While the
experiments show an increase in deposition thickness with
axial distance, the model predicts the opposite trend. At steady
state, reasonable quantitative agreement and trends are
obtained.
Presently, more elaborate numerical models are being
developed, with the objective of trying to improve the level of
agreement with experiments. It is possible, however, that the
discrepancy observed may also be due to the influence of
deposition mechanisms other than molecular diffusion.
Concluding Remarks
The present paper presents results of an ongoing research
program aimed at studying the mechanisms responsible for
wax deposition in petroleum pipelines. The strategy employed
in the study was to conduct laboratory experiments in simple
geometries with well-defined boundary and initial conditions
employing fluids with known thermo physical properties. In
parallel with the experiments, numerical models of the
deposition process were developed incorporating molecular
diffusion as the only deposition mechanism. A comparison of
the experimentally and numerically predicted results may help
assess the validity of each deposition mechanism.
Experiments were conducted for a stagnant layer of fluid
sealed in a cavity and submitted to a transverse temperature
gradient. Glass walls allowed visual access to the deposition
process. Close up images of the deposition process were
obtained with the help of a microscope. The images revealed
that, besides the wax crystal formation at the cool wall, wax
crystal are also formed at the vicinity of the wall and are freely
moving before being trapped by the network formed in the
later stages of the deposition process.
Experiments for the wax deposition in laminar channel
flow were also conducted. Again close up images of the
deposition process allowed, seemingly for the first time, the
measurement of the spatial and temporal evolution of the
deposited layer.
The diffusion-based model developed to simulate wax
deposition in the stagnant fluid was able to predict the correct
trend of the evolution of the deposition front. However, the
model significantly under-predicts the deposition thickness,
indicating that other deposition mechanisms may be present.
In the case of deposition under laminar flow, the
molecular-diffusion-based numerical model displayed a poor
predicting capability. Deposition thickness did not present
even the correct trends, with the exception of the steady state
configuration.
It is believed that the strategy adopted in the present
research may contribute to a better understanding of the
relative importance of the wax deposition mechanisms. The
data generated is of good quality and was produced under
well-defined conditions, being a good database for model
testing. These data however are based on simple laboratory
conditions and test fluids that are far from the actual field
conditions.
Acknowledgement
The authors acknowledge the support awarded to the
present work by the Petrobras R&D Center and to CNPq, the
Brazilian Research Council. The authors thank Prof. A.O.
Nieckele and M.I. Romero from the ME Department of PUC-
Rio for providing the numerical computations for the laminar
channel flow tests.
References
[1] Burger, E.D., Perkins, T.K., and Striegler, J.H., 1981, Studies of
Wax Deposition in the Trans Alaska Pipeline, Journal of
Petroleum Technology, June, pp. 1075-1086.
[2] Brown, T.S., Niesen, V.G., and Erickson, D.D., 1993,
Measurement and Prediction of the Kinetics of Paraffin
Deposition, 68th Annual Conference of the Society of Petroleum
Engineers, paper no. SPE 26548.
[3] Creek, J.L., Lund, H.J., Brill, J.P., and Volk, M., 1999, Wax
Deposition in Single Phase Flow, Fluid Phase Equilibria,
Vol.158-160, pp. 801-811.
[4] Svendsen, J.A., 1993, Mathematical Modeling of Wax
Deposition in Oil Pipeline Systems, AIChE Journal, Vol.39,
No.8, pp. 1377-1388.
[5] Azevedo, L.F.A. and Teixeira, A.M., 2003, A Critical Review
of the Modeling of Wax Deposition Mechanisms, Petroleum
Science and Technology, Vol.21, No.3 and 4, pp. 393-408.
[6] Ribeiro, F.S., Souza Mendes, P.R., and Braga, S.L., 1997,
Obstruction of Pipelines due to Paraffin Deposition during the
Flow of Crude Oils, Int. J. Heat Mass Transfer, Vol.40, pp.
4319-4328.
[7] Fusi, L., 2003, On the Stationary Flow of a Waxy Crude Oil
With Deposition Mechanisms, Nonlinear Analysis, vol 53, pp.
507-526.
[8] Leiroz, A.T., 2004, Study of Wax Deposition in Petroleum
Pipelines, Ph.D. thesis, Pontifcia Universidade Catlica do Rio
de Janeiro PUC-Rio, Rio de Janeiro, Brazil (in portuguese).
OTC 17081 7

Figure 1(a) and (b) - Schematic views of the test section constructed for visualization of wax deposition. (dimensions in mm).

Figure 2(a) and (b) - Schematic views of the test section for microscopic visualization of wax deposition. .

Figure 3 - Schematic view of the test section for wax deposition studies in channel flow.
8 OTC 17081

Figure 4 - Schematic view of the test section for wax deposition studies in channel flow

(a) (b)

(c) (d)

Figure 5 - Evolution of deposited wax within cavity. Hot and cold wall temperatures equal to 320C and 120C. External ambient
temperature equal to 25oC. Images represent intervals of 0, 15, 48, and 125 seconds after cooling starts.
OTC 17081 9

(a) (b)
Figure 6 - Visualization wax crystals formation process near the cold wall. Cold wall in the left side of the figures. Magnification of
200x. Cooling rate of 0,1oC/min. .

(a) (b)
Figure 7 - Visualization wax crystals formation process near the cold wall. Cold wall in the left side of the figures. Magnification of
1000x. Cooling rate of 0,01oC/min. .

30
T=20C

+ + +
+ + +

25
+

+
20
+

+
(mm)

15
+ Ambient Temperature (C)

+
23.5
+

21.7
10 + 21.7
+ 28.0
+
++

28.8
++ 25.1

25.3

+

25.3
5 +
+ Le = 2,96

Le = 29,6
+
+ Le = 296

0
+
0 250 500 750 1000
t (s)
Figure 8 - Measured and predicted time variation of wax deposit thickness in a stagnant cavity.
10 OTC 17081

Figure 9 - Measured deposition thickness at steady state for laminar channel flow. Inlet fluid temperature of 40oC. Wall temperature
of 15oC. Reynolds number of 580. WAT equal to 36oC.

Figure 10 Measured spatial and temporal evolution of deposited wax layer for channel flow. Inlet fluid temperature of 40 oC. Wall
temperature of 15oC. Reynolds number equal to 301. WAT equal to 36oC.
OTC 17081 11

Figure 11 - Measured spatial and temporal evolution of deposited wax layer for channel flow. Inlet fluid temperature of 40 oC. Wall
temperature of 15oC. Reynolds number equal to 729. WAT equal to 36oC

3.0
Re = 301
Deposition thickness (mm)

2.5

2.0

60 s
1.5
180 s
300 s
1.0
600 s
steady steate
0.5

0.0
0 50 100 150 200 250 300 350
axial length (mm)
Figure 12 Comparison of measured and predicted evolution of the wax deposited layer for channel flow. Inlet fluid temperature of
40 oC. Wall temperature of 15oC. Reynolds number equal to 301. WAT equal to 36oC