Energy & Fuels 2008, 22, 2693–2706 2693

Wax Deposition and Aging in Flowlines from Irreversible

Thermodynamics
Hussein Hoteit,‡ Reza Banki,†,‡ and Abbas Firoozabadi*,‡,§
Imperial College London, London SW7 2AZ, U.K., ReserVoir Engineering Research Institute (RERI),
385 Sherman AVenue Suite 5, Palo Alto, California 94306, and Department of Chemical Engineering, Yale
UniVersity, 9 Hillhouse AVenue ML103, New HaVen, Connecticut 06511

ReceiVed NoVember 30, 2007. ReVised Manuscript ReceiVed April 4, 2008

The development of waxy crude oil and some gas condensate fields can lead to serious operational problems
because of solidification of the paraffin components of the fluid in flowlines. Many numerical models in the
literature predict the thickness of the wax deposit. However, most of these models assume that the wax-oil
(gel) deposit has a constant wax content. In this work, we analyze wax deposition in laminar flow regime to
predict the thickness and the composition of the gel layer as a function of position and time. The wax-oil gel
region is considered as a porous medium. The velocity field and the pressure drop are calculated from the
Navier-Stokes equation in the liquid region and from a combined Darcy-type equation and the Navier-Stokes
equation in the gel region. The wax amount is estimated as a result of a decrease in fluid temperature below
the wax appearance temperature (WAT), counterdiffusion processes from thermal and molecular diffusions,
and radial convection which occurs because of nonuniform gel layer thickness. We compare predicted results
from our model with several experimental data from the literature. The results which are in agreement with
data cannot be predicted by formulations in which chain rule is used to replace concentration gradient with
temperature gradient in the molecular diffusion expression.

1. Introduction settling.6–11 Burger et al.7 reported that molecular diffusion of

Wax consists mainly of heavy paraffins and some naphthenes.
At temperatures below the wax appearance temperature (WAT),
wax components drop out of the solution and crystallize. Wax
deposition in production tubing and pipelines is a common
problem in cases where the fluid temperature is less than the
WAT. Wax deposition along the inner walls of the pipeline
increases the pressure drop, decreases the flow rate, and causes
operational problems. To prevent blockage of pipelines, wax
deposits should be removed periodically. Different mechanical,
thermal, and chemical techniques can be used for wax removal.1–5
There is considerable interest in predicting the deposition rate,
wax thickness, and wax content as this helps to its prevention
and use of various methods for its removal.
A large number of studies have been conducted on various
aspects of wax deposition in pipelines. Mechanisms considered
in these studies include the following: (1) molecular diffusion,
(2) shear dispersion, (3) Brownian diffusion, and (4) gravity
* Corresponding author. E-mail: abbas.firoozabadi@yale.edu.
† Imperial College London.
‡ RERI.
§ Yale University.
(1) Shock, D.; Sudburg, J. D.; Crockett, J. J. Studies of the mechanism
of paraffin deposition and its control. J. Pet. Technol. 1955, 7 (9), 23–30.
(2) Jorda, R. M. Paraffin deposition and prevention in oil wells. J. Pet.
Technol. Trans. AIME. 1966, 237, 1605–1612.
(3) Narvaez, C.; Ferrer, A. A.; Corpoven, S. A. Prevention of paraffin
well plugging by Plunger-Lift use, SPE 21640. Presented at The SPE
Production Operation Symposium, Oklahoma City, OK, April 7-9, 1991.
(4) Svetgoff, J. Paraffin problems can be resolved with chemicals. Oil
Gas J. 1984, 82 (9), 79–82.
(5) Eastund, B. J.; Schmitt, K. J.; Meek, D. L.; Anderson, D. C.;
Grisham, G. New system stops paraffin buildup. Pet. Eng. Int. 1989, 61
(1), 46–51.
species is a dominant process at high temperatures and high
heat flux conditions whereas shear dispersion is the dominant
flux at low temperatures and low heat fluxes. Also, the
contribution of the Brownian motion is small compared with
other mechanisms. Hamouda and Ravnoy9 concluded that
molecular diffusion is the main mechanism. Brown et al.12
concluded that increasing shear rates (flow rates) decreases the
rate of deposition. Hamouda and Davidsen13 investigated the
effect of flow rate on wax deposition. Their studies reveal a
significant change of the wax deposition rate when the flow
(6) Bern, P. A.; Withers, V. R.; Cairns, J. R. Wax Deposition in Crude
Oil Pipelines. Proceedings of the European Offshore Petroleum Conference,
London, England, Oct 21-24, 1980; exhibit paper EUR 206.
(7) Burger, E. D.; Perkins, T. K.; Striegler, J. H. Studies of Wax
Deposition in the Trans Alaska Pipeline. J. Pet. Technol. 1981, 33 (6), 1075–
1086.
(8) Majeed, A.; Bringedal, B.; Overa, S. Model Calculates Wax
Deposition for N. Sea Oils. Oil Gas J. 1990, 88 (25), 63–69.
(9) Hamouda, A. A.; Ravnoy, J. M. Prediction of Wax Deposition in
Pipelines and Field Experience on the Influence of Wax on Drag-Reducer
Performance. 24th Annual Offshore Technical Conference, OTC 7060,
Houston, TX, May 4-7, 1992.
(10) Rygg, O. B.; Rydahl, A. K.; Ronningsen, H. P. Wax deposition in
Offshore Pipeline Systems. 1st North American Conference, June 1998.
(11) Weingarten, J. S.; Euchner, J. A. Methods for Predicting Wax
precipitation and Deposition, SPE 15654-PA. SPE Prod. Eng. 1988, 3 (1),
121–126.
(12) Brown, T. S.; Niesen, V. G.; Erickson, D. D. Measurement and
Prediction of the Kinetics of Paraffin Deposition. SPE Annual Technical
Conference and Exhibition, Houston, TX, Oct 3-6, 1993; SPE 26548-MS.
(13) Hamouda, A. A.; Davidsen, S. An Approach for Simulation of
Paraffin Deposition in Pipelines as a Function of Flow Characteristics with
Reference to Tesside Oil Pipeline. Proceedings of the SPE International
Symposium on Oilfield Chemistry, San Antonio, TX, Feb 14-17, 1995;
SPE 28966-MS.

10.1021/ef800129t CCC: $40.75  2008 American Chemical Society

Published on Web 06/18/2008
2694 Energy & Fuels, Vol. 22, No. 4, 2008 Hoteit et al.

Table 1. Relevant Data for Example 1

parameter value
L (m) 0.254
Rin (mm) 1.841
Rout (mm) 3.175
znC8 0.67
zcycloC6C19 0.33
Tin (°C) 0
Ta (°C) 25
Q (gpm) 0.0158 (10-6 m3/s)
Re 526
knC8 (W/m · K) 0.13
kcycloC6C19 (W/m · K) 0.19
kglass (W/m · K) (pipe wall) 17.4
FcycloC6C19 (kg/m3) at 25 °C 827
FnC8 (kg/m3) at 25 °C 699
Fglass (kg/m3) at 25 °C 2230 Figure 1. Computational domain and definition of the two main regions.

species Tc (K) Pc (bar) acentric factor MW (g/mol)

nC8 568.7 24.92 0.39 114.0
cycloC6C19 840.0 10.86 0.94 350.0

shifts from laminar to turbulent flow. Creek et al.14 concluded

that the deposition rate decreases with increasing flow rate rather
than increasing as suggested by a number of authors. They also
concluded that the liquid fraction of the gel layer in turbulent
flow is significantly lower than that in laminar flow. Several
authors have argued that shear dispersion, Brownian diffusion,
and gravity settling can be neglected in the deposition of
wax.12–16 Molecular diffusion and heat transfer are often
suggested to be the main cause of deposition.12,13,15
The numerical simulation of wax deposition in pipelines is a
difficult task. The numerical model should take into consider-
ation the thermodynamic nature of the problem and the moving
boundary between the gel layer and the liquid bulk. The
thermodynamic modeling for solid/liquid equilibrium has been
successfully studied by many authors.17,18 However, modeling
fluid dynamics, thermal conduction, and multicomponent dif-
fusions owing to molecular and thermal gradients in the wax-
oil gel layer and the liquid region are a challenge. Most of the
numerical models in the literature are based on some assump-
tions which may not be valid. Sevendsen19 is perhaps the first
author who introduced a mathematical model for quantitative
prediction of wax components in the deposit for a horizontal
pipe. The model uses simplified fluid dynamic correlations to
predict the velocity field and the temperature profile. The
velocity is assumed essentially axial which is calculated from
a power-law correlation that depends on the radius of the wax-
deposit layer. This correlation may not be accurate if there is a
significant variation in the thickness of the deposit along the
pipeline. Sevendsen ignored the horizontal temperature gradient
which can only be justified for long enough pipes. The model
is based on the assumption that the wax content is constant in
the gel layer. However, experiments have shown that wax
(14) Creek, J. L.; Matzain, B.; Apte, M.; Volk, M.; Delle Case, E.; Lund,
H Mechanisms for Wax Deposition. AIChE Spring National Meeting,
Houston, TX, March 1999.
(15) Singh, P.; Fogler, H. S.; Venkatsean, R.; Nagarajan, N. Formation
and aging of Incipient Thin film Wax-oil Gels. AIChE J. 2000, 46 (5), 1059–
1074.
(16) Singh, P.; Fogler, H. S.; Venkatesan, R. Morphological evolution
of thick wax deposits during aging. AIChE J. 2001, 47 (1), 6–18.
(17) Pedersen, K. S. Prediction of cloud point temperature and amount
of wax precipitation. SPE Prod. Facil. 1995, 10 (1), 46–49.
(18) Lira-Galeana, C.; Firoozabadi, A.; Prausnitz, J. M. Thermodynamics
of Wax Precipitation in Petroleum Mixtures. AIChE J. 1996, 42 (1), 239–
248.
(19) Svendsen, J. A. Mathematical Modeling of Wax Deposition in oil
Pipeline Systems. AIChE J. 1993, 39 (8), 1377–1388.
Figure 2. Deposit thickness ratio vs time; comparison with repeated
measured data of Cordoba and Schall.27 Example 1.
content may vary significantly in the axial and radial directions
in the gel layer.16 Greenhill et al.20 used a semiempirical wax
model which includes the shear removal mechanisms. The
model predicts the wax volume but disregards the wax content.
Singh et al.15,16 developed a mathematical model to describe
the wax formation and aging in a laboratory flow-loop apparatus.
Based on experimental observations, their model takes into
consideration the radial variation of wax content in a thick wax-
deposit and assume constant content in a thin wax-deposit. The
velocity field and temperature profile are calculated based on
assumptions similar to Sevendsen.19 Ramirez-Jaramillo et al.21
presented a model for wax deposition in oil pipelines that
incorporates molecular diffusion, shear removal, and aging of
the gel layer. The authors used the multisolid model from Lira-
Galeana et al.18 to describe the solid/liquid phase behavior. They
assumed a quasi-steady state for mass, momentum, and energy
balances and assumed incompressible fluids. Hernandez et al.22
introduced a kinetic resistance factor in a wax model to allow
for the departure from the bulk phase equilibrium at the
interface. They also incorporated the effect of shear stripping
in their work. Assumptions in the model include the following:
(1) constant wax content in the deposit layer and (2) a single
(20) Elphingston, G. M.; Greenhill, K. L.; Hsu, J. J. C. Modeling of
Multiphase Wax Deposition. J. Energy Res. Technol. 1999, 121 (2), 81–
85.
(21) Ramirez-Jaramillo, E.; Lira-Galeana, C.; Manero, O. Modeling Wax
Deposition in Pipelines. Pet. Sci. Technol. 2004, 22 (7-8), 821–861.
(22) Hernandez, O. C.; Hensley, H.; Sarica, C.; Brill, J. P.; Volk, S. P. E.;
Delle-Case, E. Improvements in Single-Phase Paraffin Deposition Modeling,
SPE 84502-PA. SPE Prod. Facil. 2004, 19 (4), 237–244.
Wax Deposition and Aging in Flowlines
Figure 3. Predicted solid saturation (fraction) and temperature at different times: Re ) 526, example 1.
diffusion coefficient for multicomponent condensate and crude
system. Other aspects of the work by Hernandez et al. are similar
to the work of Singh et al.15
In all existing models, the convection flow in the gel layer is
neglected and Fick’s law is used to describe the molecular
diffusion. The convection flow can be important during the early
formation of the gel layer. It is reported that as little as 2%
weight fraction of precipitated wax is sufficient to form the gel
layer.16,23,24 The gel layer, with significant amount of oil trapped
in it, behaves as a pseudoporous medium with nonzero convec-
tion flow.7 To the best of our knowledge, none of the available
studies in the literature have accounted for the radial convection
flux in the wax formation. There are two concerns about the
use of Fick’s law in describing the molecular and thermal
diffusions. The description of mass diffusion by Fick’s law is
only valid for isothermal process. Fick’s law may not directly
apply to describe diffusion processes in the gel layer or in the
liquid region because of the radial temperature variation. The
second concern is that the available correlations for predicting
diffusion coefficients are only appropriate for binary mixtures.25
Most of the available models that support multicomponent flow
(23) Holder, G. A.; Winkler, J. Wax Crystallization from Distillate Fuels:
I. Cloud and pour Phenomena Exhibited by Solutions of Binary n-paraffin
Mixtures. J. Inst. Pet. 1965, 51 (499), 228–235.
(24) Holder, G. A.; Winkler, J. Wax Crystallization from Distillate Fuels,
Part II: Mechanisms of Pour Depression. J. Inst. Pet. 1965, 51 (499), 235–
243.
Energy & Fuels, Vol. 22, No. 4, 2008 2695
use one average diffusion coefficient by treating the oil solvent
components and the wax components as a binary mixture.7,15,16,19
Singh et al.15,16 suggested that the aging process is because of
a counter-diffusion process where the heavy components diffuse
into the gel layer and the light components diffuse in the
counterdirection. The carbon number at which the oil and wax
molecules segregate and counterdiffuse is called as the critical
carbon number by Singh et al.15,16 The critical carbon number
could be different for different compositions, and is also a
function of the wax precipitation conditions, like the pipe wall
temperature.15,16 Models that simplify the diffusion processes
by using an average molecular diffusion coefficient may have
a drawback because they cannot predict the critical carbon
number. Furthermore, the critical carbon number cannot be
predetermined. The appropriate diffusion flux in flowlines is
the result of both molecular and thermal diffusion processes.14,25
In a recent work,26 we have developed a mathematical model
for wax deposition in pipelines that has two basic differences from
the existing models. First, in this model, the diffusion processes
are described by the molecular diffusion and the thermal diffusion.
The molecular and thermal diffusions are driven by compositional
and thermal gradients, respectively. The thermal and molecular
diffusion coefficients of components vary with position and time.
(25) Ghorayeb, K.; Firoozabadi, A. Molecular, Pressure, and Thermal
Diffusion in Nonideal Multicomponent Mixtures. AIChE J. 2000, 46 (5),
883–891.
2696 Energy & Fuels, Vol. 22, No. 4, 2008 Hoteit et al.

Figure 4. Predicted temperature, solid saturation, and wax composition variations: (a) temperature at the pipe midsection vs radial distance, (b)
temperature averaged over the radius vs axial distance; (c) solid saturations at two points A and B; (d) composition of nC8 at two points A and B:
Re ) 526, example 1.

Table 2. Relevant Data for Example 2a

parameter 0.67 wt % wax 3 wt % wax
L (m) 2.44 2.44
Rin (mm) 7.2 11
Rout (mm) 8 11.8
wax type Mobil M140 Mobil M140
solvent type 3:1 Blandol and kerosene 3:1 Blandol and
kerosene
wax amount (wt %) 0.67 3
Tin (°C) 22.2 29.5
Ta (°C) 4.4; 7.2; 8.3 4.5; 9.5; 14.5
Q (gpm) 1; 2.5; 4 1; 2; 4
Re 540; 1340; 2140 260; 520; 1040
Ffluid (kg/m3) at 25 °C 838.5 832.5
µfluid (mPa · s) at 25 °C 8.7 11.7
a Mixture with 0.67 and 3 wt % wax.

They can be computed from the model of Ghorayeb and Firooza-

badi25 which is based on irreversible thermodynamics. Second, the
convection flow of the liquid fluid in the gel layer is taken into
account. The convection process may have an important effect on
the deposition rate during the early formation of wax. The gel layer
is considered as a pseudoporous medium where a Darcy-type
equation is used to describe the relation between the pressure drop
and the fluid flow velocity.
Figure 5. Solubility of the wax-forming species in the oil solvent;
comparison between our predictions and measured data from Singh et
al.,15,16 example 2.
In this work, the reliability of our model from ref 26 is
validated using different experimental data from Cordoba and
Schall27 and Singh et al.15,16 The proposed model can predict
the wax thickness and aging in the gel layer as a function of
Wax Deposition and Aging in Flowlines Energy & Fuels, Vol. 22, No. 4, 2008 2697

Table 3. Critical Properties, Acentric Factor, and Molecular stops decreasing, wax continues to grow within the gel layer
Weight of the Oil and Wax Components, Example 2 mainly as a result of diffusion mechanisms. Our mathematical
species Tc (K) Pc (bar) acentric factor MW (g/mol) model for wax deposition in the laminar regime is based on the
C11 638.8 19.48 0.53 156.0 assumptions that the shear removal of the deposited wax
C12 658.4 18.10 0.57 170.0 particles, Brownian motion, and gravity settling are negligible.
C13 675.9 16.79 0.61 184.0 There is a belief by many authors that these mechanisms are
C14 692.3 15.73 0.65 198.3
C15 707.8 14.79 0.69 212.3
not significant in wax deposition9,12,14,15,28 and the molecular
Blandol and kerosene 670.8 6.23 1.80 500.0 diffusion is the dominating process. Neglecting the gravity
C22-C26 808.8 8.79 1.18 353.9 effects, the 3D physical problem is simplified to a 2D
C27-C30 832.6 7.72 1.21 403.6 computational problem, where computations are preformed in
C31-C34 852.8 6.77 1.35 454.8
C35+ 868.6 5.99 1.37 502.5
a cross section with dimensions equal to the length and the inner
radius of the pipe. In this work, we consider the gel layer as a
pseudoporous medium, where the precipitated solid phase is
position and time. It can capture the effect of flow rate and pipe immobile. Observations have shown that as little as 2% weight
wall temperature on wax formation. fraction of precipitated wax is enough to form the gel layer.15,23,24
This paper is organized as follows. In the next section, the The species in the liquid phase continue to flow into the gel
model for wax deposition is briefly reviewed. We then show layer because of convection and diffusion processes. During
the results for a two-component mixture where the effect of the early formation of the gel layer, the convection process could
flow rate and diffusion fluxes are discussed. Our model is then be dominant.
used to predict and analyze the results from different experi-
In the proposed wax deposition model, we incorporate
ments. The paper is concluded with some remarks.
momentum, energy, and species balance equations and a
thermodynamic model for liquid/solid phase equilibria. We
2. Wax Deposition Model
couple the Darcy law and the Navier-Stokes equation to
The gel layer forms when the fluid temperature falls below approximate the pressure and the velocity field in the gel layer
the WAT. When the gel layer forms and the fluid temperature and in the liquid region. We solve the transient Navier-Stokes

Figure 6. Wax deposition as a function of time for different outside wall temperatures: (a) wax thickness ratio; (b-d) amount of wax species. Q
) 1 gpm, initial Re ) 540, initial wax content ) 0.67 wt %, example 2.
2698 Energy & Fuels, Vol. 22, No. 4, 2008
Figure 7. Predicted solid saturation and composition vs radial distance
at the pipe midsection for three different wall temperatures: (a) solid
saturation, (b) composition of Blandol, and (c) overall C35+ composition.
Initial wax content ) 0.67 wt %, time ) 5 days, example 2.
equation for compressible fluid in the liquid region. In the gel
layer, a source term of Darcy-type is added to the momentum
equation. This source term aims to slow down the velocity as
the amount of solid wax increases in the gel layer. The volume
fraction of the liquid phase (liquid saturation) can be viewed
as the porosity of the porous medium. The permeability of the
Hoteit et al.
Figure 8. Predicted temperature and velocities vs radial distance at
the pipe midsection for three different wall temperatures: (a) temperature
and (b) axial and (c) redial velocities. Initial wax content ) 0.67 wt
%, time ) 5 days, example 2.
porous medium is written as a function of porosity by using
the Carman-Koseny equation.29 The energy balance equation
is written in terms of the enthalpies of liquid and solid phases.
The Peng-Robinson equation of state (PR-EOS) is used to relate
the enthalpy of the liquid phase to the temperature. The enthalpy
Wax Deposition and Aging in Flowlines
Figure 9. Driving forces for molecular and thermal diffusions vs radial
distance at the pipe midsection: (a) composition gradient of C35+ in
liquid phase; (b) temperature gradient. Initial wax content ) 0.67 wt
%, time ) 5 days, Ta ) 4.4 °C, example 2.
of the solid phase is calculated in terms of the molar latent heat,
heat capacity, and enthalpy of pure components.30 In the species
balance equations, the convection and diffusion processes are
taken into consideration in the gel layer and in the liquid region.
The diffusion flux is written as a combination of the molecular
and thermal diffusions. The model of Ghorayeb and Firooza-
badi25 is used to compute molecular and thermal diffusion
coefficients. For a given temperature, pressure and composition,
the multisolid wax model by Lira-Galeana et al.18 is used for
solid/liquid phase split calculation. This model is also used by
Ramirez-Jaramillo21 for a similar purpose. From our experience,
this model is reliable. The successive substitution iteration (SSI)
technique introduced in Banki et al.26 is numerically very
efficient. Unlike most wax deposition models, we do not assume
a quasi-steady state for the momentum and energy balance
equations. All equations are approximated in the transient state.
This allows the use of our model for shut down conditions.
A finite-difference based method is used for the spatial
discretization of the 2D computational domain. The algorithm
starts by approximating the pressure and the velocity field. The
SIMPLER method of Patankar31 is used to approximate the
momentum equation that is coupled with the overall mass
balance equation. After calculating the pressure and the velocity
field, the temperature and the composition are predicted by
Energy & Fuels, Vol. 22, No. 4, 2008 2699
Figure 10. (a) Wax thickness ratio as a function of time for different
wall temperatures. (b) Gel layer profiles in the computational domain
for different wall temperatures: Q ) 1 gpm, initial Re ) 260, time )
4 days, initial wax content ) 3 wt %, example 2.
solving the energy and species balance equations. For a given
temperature, pressure and composition, the phase-equilibria
model is applied to calculate solid and liquid amounts and the
densities. We refer to Banki et al.26 for a complete description
of the mathematical model and the algorithm.
3. Results and Discussions
We use our model to predict deposition in the laboratory
setups from Cordoba and Schall27,32 and the comprehensive work
of Singh et al.15,16 Our predictions are compared with the
measurements by these authors. We also report on the effect of
molecular and thermal diffusion on the deposition. Then we
present calculated results for a long pipe.
(26) Banki, R.; Hoteit, H.; Firoozabadi, A. Int. J. Heat Mass Transfer,
published online Feb 7, http://dx.doi.org/10.1016/j.ijheatmasstrnasfer.
2007.11.012.
(27) Cordoba, A. J.; Schall, C. A. Solvent migration in a paraffin deposit.
Fuel 2001, 80, 1279–1284.
(28) Chen, X. T.; Butler, T.; Volk, M.; Brill, J. P. Techniques for
measuring wax thickness during single and multiphase flow. SPE Annual
Conference and Exhibition, San Antonio, TX, Oct 5-8, 1997; SPE 38773-
MS.
(29) Carman, P. C. Trans. Inst. Chem. Eng. 1937, 15a, 150–166.
(30) Firoozabadi, A. Thermodynamics of hydrocarbon ReserVoirs;
McGraw-Hill: New York, 1999.
2700 Energy & Fuels, Vol. 22, No. 4, 2008
Figure 11. Wax deposition as a function of time for different flow
rates: (a) wax thickness ratio; (b and c) amount of wax species. Initial
wax content ) 0.67 wt %, example 2.
The numerical model requires, as input data, the pipe
diameter, fluid composition and critical properties, thermal
conductivities of pure components in the liquid state, and
(31) Patankar, S. V. Numerical Heat Transfer and Fluid Flow; Hemi-
sphere Publishing Corp.: Washington, 1980.
Hoteit et al.
thermal conductivities of the solid-wax components. The
melting-point properties, heat capacity, and enthalpy of fusion
for paraffins, naphthenes, and aromatics can be calculated by
using existing correlations.25 Heat capacities and molar latent
heat of pure components can be found in Chickos et al.33 and
Himrad et al.34
3.1. Example 1: Comparison with Data from Cordoba
and Schall. In this example, we compare our predictions with
experimental data by Cordoba and Schall;27 these authors
measured the wax deposition thickness in a flow loop system.
The fluid mixture consists of a light oil species (n-octane) and
a heavy cycloalkane (nonadecylcyclohexane; cycloC6C19). Wax
formation is investigated as a function of time in a 25.4-cm
testing tube. The fluid is recycled in the flow loop at a flow
rate of 63 mL/min, which corresponds to a clean-tube Reynolds
number of 526. The wall thickness of the testing tube is about
1.3 mm, and the inner tubing radius is 1.8 mm. Therefore, the
pipe wall resistance to heat transfer cannot be neglected. We
calculated the transient solution of the energy balance equation
in the computational domain which includes the pipe wall. The
flow conditions and other relevant data are presented in Table
1. Cordoba and Schall32 evaluated the amount of the deposition
by using two methods: (1) the heat transfer method and (2) the
gravimetric method. They compared the two methods and
concluded that the former is more precise than the latter,
especially with thick deposition. They explained that the error
in the gravimetric method is due to the difficulty in draining
the trapped oil from the collected gel sample. Our predictions
are thus compared with the results reported based on the heat
transfer method. Cordoba and Schall32 reported that 0.08 mol
fraction of cycloC6C19 is soluble in nC8 at 17 °C. In our work,
we use the measured solubility data to adjust the critical
properties of the heavy component. The parameters used in the
phase-split calculations are listed in Table 1. The inlet fluid
temperature was fixed in the 25-28 °C range and outside
wall temperature was held at 0 °C. As a result, there is a high
temperature gradient in the system. We first used one of the
experiments to calculate the parameter that represents the
morphology of the porous media comprised of wax crystallites
(see ref 26). The value of C ) 106 1/m2 describes wax
morphology. In all the other experiments in this example and
in the rest of the paper, this value will be used. The computa-
tional domain is discretized into a 35 × 29 grid. The time step
is adaptively controlled by using a simple algorithm that allows
for limited variations of the pressure and composition during
each time step.
Figure 2 presents the results for the deposition thickness ratio
δ/Rin (δ is the average deposit thickness and Rin is the pipe inner
radius) at a flow rate of 63 mL/min. Our predictions and
measured data show good agreement. The figure shows that our
model can capture (1) the rapid formation of the gel region
which occurs during the first few minutes of the experiment
and (2) the increase in wax thickness, where the deposit
thickness reaches an asymptotic value after about 60 min. The
solid saturation in the computational domain is presented in
Figure 3a-d at t ) 1, 10, 40, and 120 min. Note the evolution
(32) Cordoba, A. J.; Schall, C. A. Application of a heat transfer method
to determine wax deposition in a hydrocarbon binary mixture. Fuel 2001,
80 (9), 1285–1291.
(33) Chickos, S.; Hosseini, S.; Hesse, D. G.; Liebman, J. F. Heat
Capacity Corrections to a Standard State: A Comparison of New and Some
Literature Methods For Organic Liquids and Solids. Struct. Chem. 1993, 4
(4), 271–278.
(34) Himrad, S.; Sunwono, A.; Mansoori, G. A. Characterization of
Alkanes and Paraffin Waxes for Application as Phase Change Energy
Storage Medium. Energy Sources 1994, 16 (1), 117–128.
Wax Deposition and Aging in Flowlines
Figure 12. Predicted temperature, solid saturation, and wax composition radial distance at the pipe midsection for two flow rates: (a) temperature,
(b) solid saturation, (c) composition (overall) of Blandol, and (d) C35+. Initial wax content ) 0.67 wt %, time ) 5 days, example 2.
of the saturation profiles and the hardening (aging) of the gel
layer. Figure 3e and 3f shows the temperature profiles at t ) 1
and 120 min. Figure 4a depicts the radial variation in the
temperature at the middle of the pipe at t ) 1, 10, and 120
min. The temperature difference is nearly 6 °C across the inner
radius of the tube at t ) 120 min. Figure 4b shows the
temperature averaged over radius along the tube length. The
predicted temperature difference between the inlet and the outlet
after 2 h is about 3 °C, which is in agreement with the
measurements reported by Cordoba and Schall.27 Figure 4c and
4d depicts the evolution of the solid saturation and the overall
composition of n-octane as a function of time at positions A
and B (see Figure 1). It can be noticed that solid saturation
(that is, wax amount) increases very quickly close to the wall,
which is mainly because of a large thermal flux. The light
component (n-octane) diffuses out of the gel region and the wax
component (cycloC6C19) moves toward the pipe wall (Figure
4d).
3.2. Example 2: Comparison with Data from Singh et
al.15,16. Singh et al.15,16 have conducted extensive measurements
on deposition from several experiments. They investigated the
influence of the operating conditions on the formation and the
aging of the gel region. They also presented a numerical model
to predict the measured data. Depending on the thickness of
the gel layer, they distinguished between two types of wax
deposition models. In one case, where the gel layer thickness
is up to 20% of the inner pipe radius (thin gel), they assume a
Energy & Fuels, Vol. 22, No. 4, 2008 2701
constant wax content in the gel region in the radial direction.15
The other case corresponds to a thick wax deposit, where the
thickness of the gel layer is up to 50% of the inner radius.16
Their model, in this case, accounts for the variation of wax
content along the radial distance. Singh et al.15,16 adjusted the
aspect ratio of the wax crystals to match the results from their
experiments and their model. They assumed a linear variation
of average aspect ratio with wax content of the deposit resulting
in changes from 1 to 24 in the aspect ratio in some of their
experiments.
Singh et al.15,16 report correctly that the aging process is due
to diffusion but only molecular diffusion is singled out as the
dominant process. Their work is based on the definition of a
critical carbon number for the wax-forming molecules. Hydro-
carbons that have carbon numbers more than the critical carbon
number diffuse into the gel layer and those with carbon numbers
less than the critical carbon number diffuse out of the gel layer
toward the pipe center. The critical carbon number is allowed
to be a function of compositions and the pipe wall temperature;15,16
that is obtained from adjusting their model to the measurements.
Our model for diffusion flux is free of a critical carbon number.
Singh et al.15,16 use the effective molecular diffusion coefficient
in a binary mixture from the critical carbon number; this
coefficient is assumed to be related to the square of the aspect
ratio.15,16 A correlation from the literature is used for binaries
by these authors.
2702 Energy & Fuels, Vol. 22, No. 4, 2008
Figure 13. (a) Wax thickness ratio as a function of time for different
flow rates. (b and c) Effect of molecular and thermal diffusions on
deposition thickness and wax aging. Initial wax content ) 3 wt %, Q
) 2 gpm, example 2.
In this example, we used our model without parameter
adjustment in the deposition and without the introduction of
the critical number. The model is applied without adjustments
for thin and thick wax depositions. As we will see later, one of
the experiments that was judged by previous authors to be a
thin wax deposition in the past, may have features of a thick
wax deposition.
In the experiments for a thin gel layer, Singh et al.15 used a
liquid solvent mixture (Blandol and kerosene) with 0.67%
Hoteit et al.
Figure 14. Radial flux of species by convection flow, molecular
diffusion, and thermal diffusion. Initial wax content ) 3 wt %, Q ) 2
gpm, example 2.
weight of wax-forming species (food grade wax). In the work
of Singh et al.,16 the same solvent mixture and wax-forming
species are used but the amount of wax-forming species was
increased to 3 wt % in order to conduct experiments for thick
deposits. Singh et al.15,16 studied the effects of the flow rate,
the pipe wall temperature on the thickness of the gel layer, and
the increase of the wax content in the deposition. The relevant
data on the testing tube and fluids and solids are presented in
Table 2. The liquid solvent (that is, oil solvent) used in the
experiments is a 3:1 mixture (volume basis) of mineral oil
(Blandol) and kerosene (see also, ref 35). The density and
viscosity of the oil solvent mixture are 838.5 kg/m3 and 8.7
mPa · s, respectively. The molecular weight distribution of the
food grade wax (Mobil M140) is given by Singh et al.,15,35 and
the carbon number varies from 23 to 38. The solubility of the
(35) Singh, P.; Fogler, H. S.; Nagarajan, N. Prediction of the Wax
Content of the Incipient Wax-Oil Gel in a Flowloop: An application of the
Controlled-Stress Rheometer. J. Rheol. 1999, 43 (6), 1437–1459.
Wax Deposition and Aging in Flowlines
Figure 15. Solid saturation and composition at the pipe midsection
for different flow rates. (a) Solid saturation. (b) Overall composition
of C35+. Initial wax content ) 3 wt %, time ) 4 days, example 2.
wax-forming species in the oil solvent is provided as a function
of temperature:35,36
T(F) 5977
Wwax ) ( 63.187 )
where Wwax is the weight percent of wax-forming species in
the mixture (solubility) and T is the temperature in degrees
Farenheit.
Figure 5 presents the measured solubility of wax-forming
species in the solvent as a function of temperature. We used
the solubility data in Figure 5 to make minor adjustments in
critical properties of the heavy species of the solid/liquid phase-
equilibria model. The critical properties of the species are
provided in Table 3 (after minor adjustments from ref 30). Note
that the adjustment of the heavy species critical properties is
not related to the flow model which is the central theme of this
work.
3.2.1. Effect of Pipe Wall Temperature. Singh et al.15
investigated the effect of the difference between the inlet fluid
temperature and the pipe wall temperature. They presented a
series of flow loop experiments where the inlet temperature is
(36) Rønningsen, H. P.; Bjørndal, B.; Hansen, A. B.; Pedersen, W. B.
Wax precipitation from north sea crude oils I. Crystallization and dissolution
temperatures, and Newtonian and non-Newtonian flow properties. Energy
Fuels 1991, 5 (6), 895–908.
Energy & Fuels, Vol. 22, No. 4, 2008 2703
kept at 22.2 °C with constant pipe wall temperatures of 4.4,
7.2, and 8.3 °C. A constant flow rate of 1 gpm (corresponding
to Re ) 540 for a clean tube) was used to perform the three
experiments. Measured data and our predictions for the wax
thickness ratio as a function of time are presented in Figure 6a;
the agreement is good. For all the three wall temperatures, one
observes the rapid formation of the gel layer to an asymptotic
value after which the gel thickness stops growing. As expected,
the gel thickness is smaller at the higher wall temperature. The
average wax content of the gel layer for the three wall
temperatures predicted from our model and the measured values
are shown in Figure 6b-d. The amount of wax continues to
increase even after the gel layer thickness stops growing. It can
be observed that wax concentration increase in the gel layer is
higher for the higher wall temperatures. To analyze this
behavior, we show, in Figures 7, the wax saturation and the
composition of Blandol (oil solvent component) and C35+ (of
wax) vs the radial distance at pipe midsection. In Figure 7a,
the wax saturation near the pipe wall is higher for the higher
wall temperature, which is consistent with Figure 6. A similar
behavior can be seen in Figure 7c for the composition of C35+.
However, the composition of Blandol has an opposite trend,
that is, lower Blandol amount for higher wall temperature. This
behavior, which is also consistent with Figures 6 and 7a, is
because of the counterdiffusion fluxes of different species that
segregate the wax components toward the pipe wall and the
solvent components away from the wall. We still need to explain
why the deposition rate is higher for thinner gel layers
(corresponding to higher wall temperature, Figure 6). Figure
8a shows that the temperature gradient close to the wall is nearly
the same for the three wall temperatures. However, in the thin
gel layer, wax components travel a shorter distance from the
bulk liquid before they accumulate at the pipe wall.
Figure 7a also shows the interesting behavior of the higher
wax saturation at the single phase liquid-gel interface than at
the pipe wall for the wall temperature of 4.4 °C, which results
in the thick gel layer. This unusual behavior has been reported
by Singh et al.16 from the measurements for the thick gel layer.
They measured a higher concentration of wax deposition at the
single phase liquid-gel interface than the pipe wall. We can
only explain such a composition profile from the balance
between thermal diffusion and the molecular diffusion and the
(1) multicomponent nature of deposition. The thermal diffusion flux
depends on the temperature gradient, which is a maximum at
the pipe wall (Figure 9b). However, the molecular diffusion
flux depends on the composition (mole fraction) gradient, which
for some of the precipitated components may be a maximum
close to the interface and for some others in the gel layer away
from the interface (Figure 9a). The effect of the molecular
diffusion can also be seen in Figure 7c where C35+ mainly
accumulates close to the interface (detailed discussion will come
later). Another process that may also contribute to the wax
growth rate is the radial convection flow. Further discussion
on molecular, thermal and convection fluxes and a more detailed
discussion will be presented later. Figure 8b and 8c shows the
axial and radial velocity at the pipe midsection for the three
wall temperatures. A thicker gel layer leads to higher velocity,
as expected.
Singh et al.16 also presented a series of flow loop experiments
using the mixture with wax-forming species of 3 wt % in the
liquid mixture (Table 2). The inlet temperature was 29.5 °C
and the pipe wall temperatures were 4.5, 9.5, and 14.5 °C. The
experiments were performed with constant flow rate of 1 gpm
(corresponding to Re ) 260). Figure 10a shows good agreement
2704 Energy & Fuels, Vol. 22, No. 4, 2008
Figure 16. Predicted composition in liquid phase of wax components vs radial distance at the pipe midsection: Q ) 1 gpm, initial wax content )
0.67 wt %, time ) 5 days, example 2.
between our predictions and measured data for the gel layer
thickness, which is presented as a function of time. The profile
of the gel layer in the computational domain is presented in
Figure 10b.
3.2.2. Effect of Flow Rate. Experiments with different flow
rates were also conducted by Singh et al.15,16 to study the effect
of the flow rate on deposition. In the first set of experiments,
where the solvent oil-wax mixture with 0.67 wt % of wax is
used, the injected fluid temperature and the pipe wall temper-
ature are held at 7.2 and 8.3 °C. Three flow rates were used: 1,
2.5, and 4 gpm (corresponding to Re ) 540, 1340, and 2140 in
Hoteit et al.
a clean tube). Since a turbulence flow could occur with the flow
rate of 4 gpm (Re ) 2140), we compared our predictions for
the wax thickness ratio and the wax content in the gel region
with the measured data at flow rates of 1 and 2.5 gpm. Figure
11 shows that the measured data and predictions are in good
agreement. Again, one notices that a higher flow rate results in
a thinner gel region with higher wax content. The same behavior
(thinner gel having higher wax content) was also observed by
varying the wall temperature. In addition to the explanation
given previously, a higher flow rate leads to a higher temperature
gradient close to the pipe wall (see Figure 12a) resulting in
Wax Deposition and Aging in Flowlines
Table 4. Relevant Data for Example 3
parameter value
L (m) 1000
Rin (cm) 15
znC8 0.67
zcycloC6C19 0.33
Tin (°C) 39
Ta (°C) 24
Q (gpm) 2 mostly from molecular diffusion (from Figure 14b). Note that
Re 500
knC8 (W/m · K) 0.13
kcycloC6C19 (W/m · K) 0.19
FcycloC6C19 (kg/m3) at 25 °C 827
FnC8 (kg/m3) at 25 °C 699
higher thermal flux. In Figure 12b-d, we show the wax
saturation and composition of Blandol and C35+ at t ) 5 days.
In another set of experiments by Singh et al.,16 where 3 wt %
wax (Table 2) was dissolved in the solvent mixture, the inlet
fluid temperature was 29.5 °C and the wall temperature was
4.5 °C. The deposit thickness was monitored as a function of
time at flow rates of 1, 2, and 4 gpm (corresponding to initial
Reynolds numbers of 260, 520, and 1040). Comparison between
predictions and measured data of the deposit thickness ratio is
presented in Figure 13a. The predictions show a slight under-
estimation of the deposit thickness as the flow rate increases.
3.2.3. Effect of Diffusion Flux and Radial ConVection
Flux. We investigated the effect of molecular and thermal
diffusions on wax composition using our numerical model for
the following cases: (1) without diffusion, (2) without thermal
diffusion (only molecular diffusion), and (3) with thermal and
molecular diffusions. Figure 13b and 13c show that molecular
and thermal diffusions have an appreciable effect on the
thickness (Figure 13b), but the effect on the aging of the gel
layer is very pronounced especially from thermal diffusion
(Figure 13c). Thermal diffusion effect has a more pronounced
effect on aging than molecular diffusion.
To appreciate the importance of the radial convection flux,
we compare the radial fluxes driven by thermal, molecular, and
convection processes across a cross section in the gel layer of
Figure 17. Predicted solid saturation and composition at different times. (a and b) Solid saturation (fraction). (c and d) Composition of cycloC6C19
(mole fraction). Re ) 500, example 3.
Energy & Fuels, Vol. 22, No. 4, 2008 2705
length L/2 and radius 0.85Rin (the location of the cross section
is shown in Figure 1). Figure 14 shows that, during the early
time of the gel layer formation, the radial convection flux is
important. However, as the solid saturation increases, the
molecular and thermal diffusions become more dominant. Figure
14a reveals that the flux of Blandol is mainly due to thermal
diffusion (see the inset); however, the flux of for the C35+ is
the plot in Figure 14a shows a reversal in molecular diffusion
flux due to sign change in composition gradient in agreement
with results in Figure 9a.
As a final presentation for the extensive tests by Singh et
al.,16 we show the computed wax and saturation and composition
of C35+ for the mixture with the high wax-forming species.
Figure 15 shows that as the flow rate increases the wax content
of the gel layer close to the interface increases substantially
above the concentration at the wall. We attribute this behavior,
which is in line with the measurements, to the multicomponent
nature of molecular diffusion and thermal diffusions. Compari-
son of the results in Figure 7a for a wall temperature of 4.4 °C,
and all the results in Figure 15 shows that one may not arbitrarily
divide wax deposition as thick or thin. The same figure also
shows a nonmonotonic wax saturation and C35+ concentration
in the radial direction. The wax saturation is close to wax
fraction in weight fraction based on the density of various
species. The results in Figure 15 are close to measured data by
Singh et al.16 However, our predicted results have different
trends from the calculated values by Singh et al.16 which show
a monotonic behavior. Figure 16 presents the predicted com-
position profile of the groupings of the wax-forming species
and may explain the nonmonotonic behavior of the composition
profiles. The figure contains the results which provide the
essence of our argument that (1) there may not exist a sound
basis to adopt a critical carbon number for deposition, (2) the
maximum concentration for different species in the gel region
may not be always at the interface, and (3) thermal diffusion
may contribute significantly to the species profiles.
2706 Energy & Fuels, Vol. 22, No. 4, 2008
3.3. Example 3: Large-Scale System. The purpose of this
example is to show the wax deposition behavior in a long
pipeline. We study the deposition of the binary mixture of nC8/
cycloC6C19 from example 1. The pipe length is 1 km, and the
inner radius is 15 cm. The flow conditions and other relevant
data are listed in Table 4. The solid saturation in the compu-
tational domain is illustrated in Figure 17a and 17b at t ) 5
and 12 months. It can be noticed that the profile of the wax
deposit in a long pipeline is different from that in laboratory-
scale tubes as shown in Figure 17a and 17b. A maximum
deposition thickness may occur far from the outlet where the
fluid temperature is the minimum. Such a behavior has been
reported by many authors.8,19–21 Figure 17c and 17d shows
composition profiles for cycloC6C19 at t ) 5 and 12 months.
The segregation of the heavy and light components is similar
to a laboratory-scale system.
4. Concluding Remarks
There are two main features that characterize our model:
We study the effect of molecular diffusion and thermal
diffusion in wax deposition predictions. Both affect the flux of
species toward the pipe wall by diffusion. The driving force
for molecular diffusion is the concentration gradient. The driving
force for thermal diffusion is temperature gradient. These two
gradients may not be proportional in the gel layer. The
temperature gradient is expected to be high close to the pipe
wall. The concentration gradient, because of phase change, is
expected to be high at the point where a species crystallizes
first in the gel layer. Furthermore, the temperature gradient close
to the wall may reach a pseudo steady state quickly and stay
large close to the wall. This may not be true for the concentration
Hoteit et al.
gradients. All these aspects may lead to incorrect results by the
commonly accepted use of the chain rule to replace concentra-
tion gradient by temperature gradient in the Fick’s law.
Our numerical model for wax deposition in petroleum
flowlines has been verified using various measurements in the
literature. The proposed model predicts the wax formation and
aging as a result of a decrease in the temperature below the
wax appearance temperature (WAT) and the continuous growth
of wax content due to thermal and molecular fluxes. Published
experimental data are used to investigate the reliability of the
model and to show that it can capture various observations in
the experiments. The predicted results show that the thickness
of the gel layer increases rapidly in the early stages, then the
deposit thickness virtually stops increasing. The wax content
of the gel layer grows as a result of the combined molecular
and thermal diffusion fluxes. It is also shown that higher flow
rates (similarly, higher wall temperatures) result in a thinner
gel layer with higher wax content. The prediction is consistent
with the measured data. We have also investigated the effect
of molecular, thermal diffusion fluxes, and the radial convection
flux. Both molecular and thermal diffusions affect the wax
deposition; the effect of thermal diffusion can be even more
pronounced than molecular diffusion. All these predictions are
made without adjustments of the parameters unlike the past work
when thermal diffusion is neglected.
Acknowledgment. The funding for this work was provided by
the member companies of the Reservoir Engineering Research
Institute (RERI). We thank Daniel Rosner and Alana Leahy-Dios
of Yale University for reading the work and their comments.
EF800129T