Wax-Deposition Forecast
Andrey Semenov, Total E&P Russie
Summary
In the case of altering the pressure/volume/temperature properties
of oil, change of the aggregate state of oil is one of the main
issues in oil processing. When oil lifts to the surface, the most ad-
vantageous energy state of high-weight components becomes the
state in which a portion of the molecules becomes solid. Phase-
behavior diagrams could be plotted using a typical compositional
simulator, but even if one knows the phase behavior, one still will
not be able to understand the wax deposition in the tubing or pipe-
line and will not forecast this process performance.
As some authors have mentioned, the resin and asphaltene con-
centrations in the oil affect the wax-deposition velocity. Because
of the high complexity of the crystallization mechanism, the high-
weight molecules are not taken into consideration in a phase-
behavior simulator. Thus, with the intent of qualitatively charac-
terizing the oil tendency to the aggregate, the oil from different
reservoirs has been compared.
Because of the temperature gradient between rock (permafrost)
and oil, there is a probability of wax deposition in the tubing. The
deposition-growth model makes different diffusion assumptions.
The equilibrium model and the film-mass-transfer model (as well
as solidification and separation) have been applied. Using these
models, the wax-growth velocity and tubing-drift diameter have
been forecast.
As a result of the harsh northern environment, there is a seri-
ous risk of oil solidification in the surface pipeline in the case of a
pump stopping. On the basis of the laboratory measurements and
with the help of a stepwise approximation, the oil-viscosity de-
pendence on temperature has been derived. By solving the heat-
transfer equation (between oil in the pipeline and the environ-
ment) and using nodal analysis, the maximum safety time in case
of an accident in the pipeline has been determined.
Phase Behavior of Crude-Oil Components
The most popular approach to characterizing the wax-appearance
tendency of oil is the development/application of phase-behavior
diagrams. Based on the equation-of-state solutions for each oil
component, phase diagrams could be derived and used to predict
solid-state concentration as a function of temperature. Wax-ap-
pearance temperature and reservoir-fluid composition are used to
calibrate and verify models. Wax-appearance temperature is deter-
mined usually as the temperature in which changes in the optical
and/or resistivity properties of oil appeared. All alkanes with mo-
lecular weight greater than a predetermined value are considered
as components of solid deposition (the model assumption) (Toms-
kNIPINeft 2007).
The solution of the preceding task for two types of oil with dif-
ferent properties is represented in Fig. 1 in asymptotic coordi-
nates. This diagram was derived using mathematical modeling of
a multicomponent system (UfaNIPINeft 2009). Wax concentra-
tion was normalized by the total wax concentration in crude oil,
and temperature was divided by the wax-appearance temperature.
Phase transition was initiated by the achievement of wax-appear-
ance temperature, and the following alteration of solution physical
properties became fixed as part of the crude oil became solid.
Wax deposition could be continued until the pour point was
achieved, at which point, oil enters the solid phase (wax
Copyright V
C 2012 Society of Petroleum Engineers
This paper (SPE 149793) was accepted for presentation at the North Africa Technical
Conference and Exhibition, Cairo, 20–22 February 2012, and revised for publication. Original
manuscript received for review 15 March 2012. Paper peer approved 20 April 2012.
November 2012 SPE Production & Operations
concentration < total wax concentration in oil). The shape of the
wax-deposition curve is determined by the crude-oil composition.
Qualitative Assessment of Wax-Appearance Risk
As mentioned in earlier papers (Mansoori 2005a,b,c), the process
of wax deposition is complicated because of the presence of resin
and asphaltene components in crude oil. The solution to the
equation-of-state system is usually simulated in the process of
deposition of alkane molecules with high molecular weight. But
analysis of wax-deposition samples from tubing has verified that
asphaltenes sometimes represent the main part of depositions
(Turukalov 2003) (the concentration of asphaltenes divided by the
concentration of resins is represented in Fig. 2 as orange for these
reservoirs). Laboratory analysis of downhole oil samples from the
fields described in the preceding displayed a high concentration of
paraffins (> 10%) and relatively small concentration of asphal-
tenes (< 1%). However, the composition of the deposits is quite
different, and the main part of the accumulated hard layer is
asphaltenes.
The various stages of the wax-deposition process can be distin-
guished as follows:
1. Crystallization of alkanes as soon as phase-transition tem-
perature has been achieved
2. Transition of the state of asphaltenes in the oil solution from
micelles (surrounded by resins) to the free state
3. Accumulation of asphaltenes at the crystallization centers
formed by the alkanes
In line with the described molecular sequence of wax deposition
in the preceding, it should be clear that eliminating the risk of
wax deposition is possible only in the case of oil temperature
greater than wax-appearance temperature in the entire oil-produc-
tion system (elimination of crystallization centers).
Mansoori (2005b) proposed to characterize the risk of wax
deposition in oil from different fields using the ratio of concentra-
tion of asphaltenes to concentration of resins. The higher the ratio,
the more serious the risk of deposition became. Stability of as-
phaltene/resin micelles is considered the main factor of deposition
appearance. In Fig. 2, the ratio was calculated for 32 different
types of oil from different fields all over the world. Actual obser-
vations of the wax-deposition process in these fields were corre-
lated to the ratio, but there are a few exceptions, with high ratio
and without hard deposition appearance.
Modeling of Wax Deposition in the Tubing
The main physical effect that determines wax growth on the inter-
nal wall of the pipe is molecular diffusion. Heat exchange
between the well and the rock is the cause of tubing cooling. As a
consequence of this process, a temperature gradient between liq-
uid flow and the wall of the tubing appears. From Fig. 1, the con-
centration of wax in oil is mainly a function of temperature. The
lower the temperature of the oil, the lower the wax concentration
in the solution (i.e., more wax in the solid state). Molecular diffu-
sion of wax molecules from the solution to the wall is caused by
the difference in wax concentration between the mixture near the
wall and that in the flow.
The mass-transfer equation for this process is (Wang et al.
2001; Hernandez et al. 2004; Bruno et al. 2008)
m_ growth ¼ m_ bulk  m_ shear  m_ aging: . . . . . . . . . . . . . . . . ð1Þ
To evaluate the coefficient of diffusion, it is possible to use either
the Wilke and Chang equation (Wilke and Chang 1955),
371
 1.01
Baxterville / USA

1 Atabasca / Canada

0.99 Alberta / Canada

0.98 Boscan / Venezuela

Cwax/Cmax

Lagrave / France
0.97
Cold Lake / Canada
0.96
Panucon / Mexico

0.95
Brookhaven / USA

0.94 Oil 1
Nadejuskoe / Russia
Oil 2

0.93 Mexia / USA

0.900 0.950 1.000 1.050 1.100
T/Tsat Hasireiskoe / Russia

Fig. 1—Wax dissolved in oil for two different crude-oil composi- Huntington Beach / USA

tions at a pressure of 10 atm based on mathematical modeling.

Cherpajuskoe / Russia

ðWMo Þ0:5 T
Tecoaminocan / Mexico

Dwo ¼ 7:4  1012 ; . . . . . . . . . . . . . . . . . . ð2Þ

lo Vw0:6 Isthmus / Mexico

or the Hayduk and Minhas approach (Hayduk and Minhas 1982), Atabasca / Brazil

10:2


0:791 Oil 1
T 1:47 lo Vw
Dwo ¼ 13:3  1012 : . . . . . . . . . . . . . .ð3Þ Tonkawa / USA
Vw0:71
Kirkuk / Iraq
In the first approximation, we will assume effects of solidification
and separation of depositions as negligible: Balachany / Russia

dd dT dC 1 Oil 2
¼ Dwo : . . . . . . . . . . . . . . . . . . . . . . . ð4Þ
dt dr dT qw Fw
Hassi Messaoud / Algeria

This model describes the equilibrium process for wax depositions.

Concentration of depositions is considered a function of tempera- Oil 2
ture only, and temperature is a continuous function of the well ra- Hould / USA
dius (true in the case of thermodynamic equilibrium), which is
dC dC dT Bibi-Eibat / Russia
described mathematically as ¼ .
dr dT dr
The equilibrium-model assumption is not fully correct in real Kaluga / Russia

conditions of wax appearance and accumulation. So, as an alter- Ok. City / USA
native, the film mass-transfer model could be used:
Davenport / USA
dd 1
¼ hm ðCbulk  Cwall Þ : . . . . . . . . . . . . . . . . . . .ð5Þ
dt qw Fw PA / USA

The wax-concentration difference between the wall and the flow Rodessa / USA

appears (because of the temperature difference), which leads to

Surachany / Russia
the diffusion process.
The mathematical models in the preceding allowed us to deter- Dossor / Russia
mine mass velocity of the wax accumulation m_ bulk , but sometimes,
a sufficient portion of molecules is solidified in the deposition 0.00 0.50 1.00 1.50 2.00

m_ shear , or separated from the deposition by the fast flow m_ aging . The
effect of wax solidification could be taken into account by intro-
ducing the efficient molecular-diffusion coefficient:
Dwo
De ¼ : . . . . . . . . . . . . . . . . . . . . . ð6Þ
ð1 þ Ka Fw Þ2 Fw
1þ
1  Fw
Later, we will consider the results of mathematical modeling
of wax growth using four models (with different hyphotheses):
 Thermodynamic equilibrium
 Film mass transfer
 Film mass transfer with solidification
 Film mass transfer with solidification and separation
Well and fluid parameters that were used for the calculations
are given in the Table 1. To model the wax concentration as a
% Asphaltenes / % Resins
Fig. 2—Ratio of average asphaltene concentration to resin con-
centration for oil samples from 32 different reservoirs. Orange
represents field with soft wax deposition in the tubing.
function of temperature, we used dependencies derived from the
simulation of the phase behavior of oil in Fig. 1.
The velocity of wax growth dd/dt is sufficiently different for
each set of model assumptions (Fig. 3). In the case of thermody-
namic equilibrium, the velocity is sufficiently slow (0.05 mm/h,)
but for the film-mass-transfer model, the velocity is faster and
decreases only in the case of the wax-separation effect, with some
asymptotic limit. The wax thickness grows dramatically, with
high velocity in the case of film mass transfer, slow velocity in

372 November 2012 SPE Production & Operations

 3
TABLE 1—WELL AND FLUID PARAMETERS
2.5

Deposition Velocity, mm/h

Parameter Value Units
2
Tubing ID 0.098 m
Tubing OD 0.114 m
1.5
Tubing length 2200 m
Tubing heat-exchange coefficient 45 W/mK 1
Fluid temperature (reservoir) 337 K
Environment temperature 273 K 0.5
Environment heat-exchange coefficient 600 W/mK
Flow rate 100 m3/d 0
Wax appearance temperature 316 K 0 5 10 15 20 25
Fluid density 823 kg/m3 Time, days

Fluid heat capacity 2037 J/kgK Equilibrium Film

Fluid heat-exchange coefficient 0.15 W/mK Film solidification Film solidification and separation
Fluid viscosity 0.7 sP
Wax density 950 kg/m3 Fig. 3—Wax-growth velocity.
Wax share in depositions (by mass) 0.5 
Wax heat-exchange coefficient 0.2 W/mK

90

80
90

Tubing Flow Area, mm

70
80
Deeposition Thickness, mm

60
70 50
60 40
50 30
40 20
30 10
20 0
10 0 5 10 15 20 25

0 Time, days
0 5 10 15 20 25
Equilibrium Film
Time, days
Film solidification Film solidification and separation
Equilibrium Film
Film solidification Film solidification and separation Fig. 5—Tubing flow area.

Fig. 4—Wax thickness.

the case of equilibrium, and even slower velocity in the case of

the wax-separation effect (Fig. 4).
Free tubing flow area for each model is displayed in Fig. 5.
20 Full tubing plugging without separation will take place after 2
18
days in the case of the film-mass-transfer model. In the equilib-
rium model, the tubing flow-area-decrease curve has a monotonic
16 shape, but in the film-mass-transfer model with separation, the
14 flow area will decrease by 12 mm in 24 days, after which effects
Shear Tension, Pa

of wax growth and wax separation will compensate each other.

12

10
Transient Modeling of Wax Deposition in the
8
Surface Pipe
6
The viscosity of oil is a function of system thermodynamic pa-
4 rameters and velocity. If viscosity is a function of temperature
only and does not depend on velocity, we can consider the fluid as
2
a Newtonian liquid. But from laboratory measurements (Fig. 6), it
0 is clear that in the low-temperature range, the shear tension is not
0 50 100 150 200 250 300 350 400 proportional to the shear velocity. Some additional shear stress is
Shear Velocity, s−1 necessary to begin oil movement.
Oil viscosity l, based on angular velocity x and shear tension
0 °C tepmerature 20 °C temperature
s in a rotational viscometer, could be calculated as

Fig. 6—Results of oil-viscosity measurements in rotational vis-

s
l¼ : . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ð7Þ
cometer. Non-Newtonian behavior at 0º C. x

November 2012 SPE Production & Operations 373

 Let us consider the problem of maximum safety time for sur-
TABLE 2—SURFACE-PIPE AND FLUID PARAMETERS face-pipe shutoff in order that the efforts of surface pumps will be
enough to begin flow. First, the model of temperature distribution
Parameter Value Units across the pipe is necessary to predict temperature after shutoff.
The parameters of the surface pipe and the fluid used in the simu-
Pipe length 172 800 m
lation are presented in Table 2. On the basis of the parameters for
Height difference amplitude 85 m
the first stage, the temperature distribution across flowing pipe
Pipe ID 0.82 m was calculated. The exponential function used as the temperature-
Pipe wall thickness 0.011 m distribution function along the working pipeline is
Thickness of isolation 0.11 m pUdo L
Isolation heat-transfer coefficient 0.035 W/mK Tout ¼ Ta þ ðTin  Ta Þe QqCp : . . . . . . . . . . . . . . . . . . ð8Þ
Steel heat-transfer coefficient 47 W/mK
Flow rate 17 000 m3/d The integral heat-transfer coefficient was estimated, taking
Fluid temperature on pipe intake 343 K into account the pumping fluid, pipe walls, isolation, and environ-
Environment temperature 273 K ment. Considering the pipe length as sufficient, we assume that
the intensity of the heat transfer between fluid and environment is
Environment heat-transfer coefficient 600 W/mK
sufficiently greater than the heat exchange inside the pipe. On the
Oil heat-transfer coefficient 0.15 W/mK
basis of this assumption, the temperature distribution along the
Oil heat capacitance 2020 J/kgK pipeline is represented in Fig. 7. Because pipe cooling causes oil
Fluid density 0.845 kg/m3 viscosity to increase and because we should consider oil as a non-
Fluid viscosity on pipe intake 2.76 sP Newtonian fluid, some additional shear stress is necessary to
begin pumping.
Oil viscosity, here, is a function not only of temperature, but also After approximating the viscosity measurements in the labora-
of shear velocity. More importantly, the oil viscosity decreases tory by means of a step function, the pipe outtake curve could be
with the increase of shear velocity. This is one of the reasons derived (Fig. 8). Then, the working point of the system (the inter-
some initial force is necessary to start oil movement in the surface section of the pump curve and the outtake curve) could be
pipe. determined.

360

340

Flow
320 1 day stop
Temperature, K

3 day stop

300 6 day stop

9 day stop
12 day stop
280

260

240
0 20 000 40 000 60 000 80 000 100 000 120 000 140 000 160 000 180 000 200 000
Pipe Length, m

Fig. 7—Temperature changes after flow shutoff.

80
Pump curve
Flow
70
12 day stop

60
Pipe Intake Pressure, atm

50

40

30

20

10

0
0 5000 10 000 15 000 20 000 25 000
Rate, m3/d

Fig. 8—Nodal analysis of pipe.

374 November 2012 SPE Production & Operations

Conclusion
A few interesting apects of the wax-deposition process were con-
sidered in the scope of this work. First, the classical approach of
liquid-/solid-phase diagrams was mentioned. The complex nature
of the wax-deposition process, which includes asphaltenes and
resins, was considered also. One of the simplest approaches to
characterizing the risk of wax appearance uses an asphaltene/resin
ratio in crude oil and depositions.
The physical mechanism of wax deposition is molecular diffu-
sion. This process is initiated by the temperature difference (and,
as a consequence, the wax-concentration difference in oil) be-
tween the main flow area and near the wall. Four different models
of the wax-deposition process were compared and discussed. The
best-match model could be selected on the basis of the typical
period of the scraping operation and could be used to optimize
well performance. This set of models could help to identify the
prevailing mechanism of depositions and, after, to optimize well
performance.
Non-Newtonian properties of oil were considered in the pro-
cess of oil flowing in the surface pipeline. Simple, nonstationary
models of pipe were developed to predict the pipe-cooling process
after shutoff. The nodal-analysis approach was used to predict the
working point (rate and pressure) in the system.
Nomenclature
Cbulk ¼ average volume of wax concentration in the flow
Cmax ¼ wax concentration dissolved in oil at the reservoir
temperature
Cp ¼ fluid heat capacity
Cwall ¼ wax concentration near the wall of the tubing
Cwax ¼ wax concentration
do ¼ pipe outside diameter
Fw ¼ wax share in depositions
Ka ¼ proportionality coefficient
L ¼ pipe length
m_ aging ¼ wax-solidification mass velocity
m_ bulk ¼ wax-volume diffusion mass velocity
m_ growth ¼ wax-growing mass velocity
m_ shear ¼ wax-separation mass velocity
Mo ¼ molecular weight of oil
Q ¼ flow rate
r ¼ radial coordinate
T ¼ temperature
Ta ¼ environment temperature
Tin ¼ fluid temperature on the intake of pipe
Tout ¼ fluid temperature on the outtake of pipe
U ¼ integral heat-transfer coefficient
Vw ¼ wax molar volume
dd/dt ¼ wax-growth linear velocity
lo ¼ oil viscosity
November 2012 SPE Production & Operations
q ¼ fluid density
qw ¼ wax density
s ¼ shear tension
W ¼ empirical diffusion coefficient
x ¼ angular velocity
References
Bruno, A., Sarica, C., Chen, H. et al. 2008. Paraffin Deposition During the
Flow of Water-in-Oil and Oil-in-Water Dispersions in Pipes. Paper
SPE 114747 presented at the SPE Annual Technical Conference and
Exhibition, Denver, 21–24 September. http://dx.doi.org/10.2118/
114747-MS.
Hayduk, W. and Minihas, B.S. 1982. Correlation for Prediction of Molec-
ular Diffusivities in Liquids. Can. J. Chem. Eng. 60 (2): 295–299.
Hernandez, O.C., Hensley, H., Sarica, C. et al. 2004. Improvements in Sin-
gle-Phase Paraffin Deposition Modeling. SPE Prod & Oper 19 (4):
237–244. SPE-84502-PA. http://dx.doi.org/10.2118/84502-PA.
Mansoori, G.A. 2005a. Asphaltene Deposition and its Control, http://www.
firp.ula.ve/archivos/material_web_4xx/05_web_Mansoori_asphaltenes.
pdf.
Mansoori, G.A. 2005b. Nanoscale Structures of Asphaltene Molecule, As-
phaltene Steric-colloid and Asphaltene Micelles and Vesicles, http://
tigger.uic.edu/mansoori/Asphaltene.Molecule_html.
Mansoori, G.A. 2005c. Paraffin/Wax and Waxy Crude Oil. The Role of
Temperature on Heavy Organics Deposition from Petroleum Fluids,
http://tigger.uic.edu/mansoori/Wax.and.Waxy.Crude_html.
TomskNIPINeft. 2007. PVT properties and core analysis of Vankor field.
R&D Report, Tomsk Oil and Gas Research and Design Institute
(TomskNIPINeft), Tomsk, Russia.
Turukalov, M.B. 2003. WAX deposition analysis on the example of Val
Gamburceva field. R&D Report, Rosneft, Krasnodar, Russia.
UfaNIPINeft. 2009. Application to the technological scheme of Vankor
field development. Technical Report, Ufa Oil and Gas Research and
Design Institute (UfaNIPINeft), Ufa, Republic of Bashkortostan, Russia.
Wang, Q., Sarica, C., and Chen, T.X. 2001. An Experimental Study on
Mechanics of Wax Removal in Pipeline. Paper SPE 71544 presented
at the SPE Annual Technical Conference and Exhibition, New Orle-
ans, 30 September–3 October. http://dx.doi.org/10.2118/71544-MS.
Wilke, C.R. and Chang, P. 1955. Correlation of diffusion coefficients in
dilute solutions. AIChE J. 1 (2): 264–270. http://dx.doi.org/10.1002/
aic.690010222.
Andrey Semenov is a reservoir engineer at Total E&P Russie.
Previously, he worked as a researcher in the Research and De-
velopment Center and as a chief specialist in the Reservoir En-
gineering Department in Rosneft. His research interests include
decision making, multiphase flow, flow in porous media, and
artificial intelligence. Semenov holds an MSc degree in
applied mathematics and physics from the Moscow Institute
of Physics and Technology.
375