SPE 87457
Barium Sulphate Oilfield Scaling: Mathematical and Laboratory Modelling
P. G. Bedrikovetsky, SPE, North Fluminense State University (LENEP/UENF); R.P. Lopes Jr., Petrobras/UN-BC; P. M.
Gladstone, North Fluminense State University (LENEP/UENF); F. F. Rosário, SPE, Petrobras/CENPES; M. C. Bezerra,
SPE, Petrobras/CENPES; E. A. Lima, Institute of Pure and Applied Mathematics (IMPA)
Copyright 2004, Society of Petroleum Engineers Inc.
This paper was prepared for presentation at the SPE 6th International Symposium on Oilfield
Scale held in Aberdeen, UK, 26-27 May 2004.
This paper was selected for presentation by an SPE Program Committee following review of
information contained in an abstract submitted by the author(s). Contents of the paper, as
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
for commercial purposes without the written consent of the Society of Petroleum Engineers is
prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300
words; illustrations may not be copied. The abstract must contain conspicuous
acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O.
Box 833836, Richardson, TX 75083-3836 U.S.A., fax 01-972-952-9435.
Abstract
The formation damage in scaled-up production wells
caused by incompatibility of injected and formation waters
have long been known. Precipitation of salts results in
permeability decline. Among the most onerous of all scaling
species is that of sulphates, particularly barium and
strontium sulphates.
We study effects of porous media on the BaSO4
scaling kinetics.
A new methodology for determination of reaction rate
coefficient from coreflood tests consists of the sequence of
diffusivity reaction free tests, of transient tests with chemical
reaction, and of simultaneous injections of both injected and
formation waters with quasi steady state concentration profiles
for barium and sulphate ions. The diffusivity tests serve for
dispersivity coefficient determination. Quasi steady state tests
allow determination of the reaction rate coefficient versus
velocity. The transient test data as compared with the
mathematical modelling data validate the model and the data
of steady state test data treatment.
An analytical model developed is used for treatment of
quasi steady state test data. The transient tests are treated by a
numerical model.
The main result of the work is proportionality between the
reaction rate coefficient and flow velocity for the parameter
range studied.
Introduction
The BaSO4 scaling is a chronicle disaster in waterflood
projects with incompatible injected and formation waters. This
is usually due to precipitation of barium sulphate from the
mixture of both waters and consequent
permeability reduction1-3.
The rate coefficient for the reaction between incompatible
chemical species in injected and formation waters is the main
parameter that determines the oilfield scaling intensity in cases
where the aqueous solution is far from equilibrium. This rate
is highly affected by flow velocity, diffusion/dispersion in
porous media, pore space geometry and, therefore, the reaction
rate coefficients inside and outside porous media should
be different.
Nevertheless, presently the reaction rate coefficients used
in mathematical modelling are obtained in laboratory reactors
without porous media4-8. Usually the solid grains are used
during water mixing in reactors in order to induce the
precipitation centres, but the pore space structure and the
relative fluid-rock flows are not represented.
The mathematical models for reactive flow in porous
media consist on mass balance equations with the reaction rate
sink terms8-10. The rate terms should depend on porous
media properties.
In order to reliably predict well behaviour during the
oilfield scaling, the effect of porous media flow on chemical
reaction rate should be studied systematically.
The design and results of barium sulphate steady state
scaling tests allowing for such study have been presented in
the literature11-13. Nevertheless, there were no attempts to
determine the reaction rate coefficients from laboratory
coreflood tests.
The laboratory and mathematical study of scaling
formation in porous media has been performed13,14. The
sequence of coreflood tests allowing for determination of the
reaction rate coefficient for various flow conditions has
been proposed:
- displacement of water by traced water at the absence
of reaction in order to determine diffusion coefficient;
- displacement of Ba-rich formation water by SO4-rich
injection water (transient tests);
- simultaneous injection of Ba-rich formation water
and SO4-rich injection water (steady state tests).
The diffusive tests allow determining the diffusion
coefficient dependency of flow velocity for a given core15. The
quasi steady state tests being performed at different velocities
determine the reaction rate coefficient versus flow velocity.
The transient tests allow comparing experimental results with
the data of numerical modelling based on the steady state
test data.
Two series of steady state tests on simultaneous injection
of formation and injection waters with barium sulphate
2 SPE 87457
formation have been performed for different velocities in two
cores. It was shown that the reaction rate coefficient is
proportional to flow velocity, in the range studied.
The fact of the velocity dependence of the reaction rate
coefficient is important in reservoir simulation, specifically in
modelling of well productivity decline during barium
sulphate scaling16.
In the current paper, an analytical model for simultaneous
injection of sea and formation waters was developed and
applied for treatment of laboratory data11-14 in order to validate
the fact of proportionality between the reaction rate coefficient
and flow velocity. The values for the proportionality
coefficient as obtained from different literature sources11-14
have the same order of magnitude.
Barium Sulphate Scaling Modelling in Porous Media
We discuss formation damage due to barium sulphate
precipitation, which is one of main physics mechanisms for
oilfield scaling productivity impairment. Usually seawater is
injected in offshore operations, and it contains SO42- anions. If
the formation water contains Ba2+ cations, mixing of injected
and formation waters may cause the BaSO4 deposition
(Fig. 1).
Ba 2 + + SO 24 − → Ba SO 4 ↓ (1)
The main assumptions of physics-mathematical model of
barium sulphate scaling are:
• the chemical reaction between barium and
sulphate ions is irreversible;
• it is a second order chemical reaction obeying the
mass action law;
• brine incompressibility;
• volume conservation during brine mixing and salt
precipitation;
• diffusion/dispersion coefficient is proportional to
flow velocity;
• independence of the rate coefficient on precipitant
concentration;
• hyperbolic form for permeability decline versus
the precipitant concentration.
The mathematical model for one-dimensional flow of injected
and formation waters in porous media, includes mass balance
equations for barium anions, for sulphate cations and for the
salt molecules, (A-1). The salt deposition rate is given by the
law of mass action 6-9,17-20, (A-2). The modified Darcy’s law
(A-3) contains the formation damage coefficient β that
describes permeability loss due to salt deposition.
The unknowns in the closed system of four equations (A-
4) are three concentrations and pressure.
The assumption that the reaction rate coefficient Ka is
independent of the precipitated salt concentration σ results in
separation of the first and second equations (A-4) from the rest
of the system.
In dimensionless co-ordinates (A-6), the system of
governing equations takes form (A-7)
⎧ ∂C ∂C ∂ 2C
+ = εD − εk C Y
⎪⎪ ∂T ∂X ∂X 2
⎨ (2)
⎪ ∂Y + ∂Y = ε D ∂ 2Y
− εk α C Y
⎪⎩ ∂T ∂X ∂X 2
where α is a ratio between the initial concentrations of Ba
in formation water and of SO4 in injected seawater.
The system (2) contains dimensionless chemical kinetics
number εk, and the number εD, which is inverse to
Peclet number15,17
0
K a c SO 4 L D
εk = ,ε D =
U LU
For the velocity range of flows in petroleum reservoirs, an
advective mass transfer highly exceeds the diffusive flux. So,
if the specie particle has already left the core, it would not
diffuse back. The corresponding outlet boundary condition is
given by (A-11)15.
The initial conditions for a reactive transient test (A-8)
correspond to the sulphate absence and to the fixed barium
concentration in porous medium before the injection.
The inlet boundary conditions (A-9) correspond to a fixed
flux of sulphate via the inlet cross section, and to the absence
of barium flux via the inlet. The assumption of negligible
diffusion simplifies the inlet conditions, as seen in (A-10).
The mathematical model (A-7)-(A-9) describes the
transient flow with displacement of formation water by
injected water accounting for irreversible chemical reaction. It
has been solved numerically by finite difference method using
Crank-Nicolson linearization21.
For quasi steady state tests, the system (2) is reduced to a
system of ordinary differential equations (C-2). The boundary
conditions (C-3),(C-4) correspond to simultaneous injection of
sea and formation waters with fixed barium and sulphate
concentrations. The boundary condition (A-11) corresponds to
the absence of diffusive return of ions that had already left the
core effluent together with the carrier water. For the case
where the sulphate concentration in seawater highly exceeds
the barium concentration in formation water (α<<1) the steady
state concentration profiles are described by explicit formulae
(C-11), (C-15)-(C-17).
Experimental Procedure
The objective of the laboratory study is the determination
of reaction rate coefficient for rock flow conditions.
The experimental setup consists on core holder with
confinement system, two pumps, and pressure transducer. The
core holder has two independent entrance tubes allowing for
simultaneous injection of two different reagents. The setup
schema and photo are given by Figs.2, 3.
The study has been carried out for scaling conditions in
field N (Brazil). The formation and seawater compositions can
be found in2,22. The formation and seawater compositions as
simulated in the current tests are presented in Table 1.
Sulphate concentration in seawater exceeds barium
concentration in formation water more than 10 times, so in the
current study α<<1.
SPE 87457
The tests have been carried out for two cores taken from
outcrop RB (Brazil). The core properties are presented in
Table 2.
The flood sequence allowing for determination of reaction
rate coefficients from the test is as following:
1. Diffusion tests:
1.1. Filling of core by simulated formation water;
1.2. Injection of simulated sea water without SO42- anions
at some velocity;
1.3. Injection of simulated formation water at another
velocity;
1.4. Further alternate injections of simulated formation
and sea waters at different velocities.
2. Transient tests:
2.1. Filling of core by simulated formation water at the
final stage of the diffusive tests;
2.2. Injection of simulated sea water at some velocity;
2.3. Injection of simulated formation water at
another velocity.
3. Quasi steady state tests:
3.1. Simultaneous injection of formation and sea waters at
some velocity;
3.2. Further simultaneous injections of formation and sea
waters at different velocities.
Diffusivity tests. The objective of diffusivity tests is
determination of diffusion/dispersion coefficient versus flow
velocity.
Diffusivity tests have been carried out without sulphate
ions. Taking off the sulphate anions from the injected water
allows avoiding chemical reaction during alternate injections
1.2-1.4 in diffusivity tests. So, barium served as a tracer.
Also, magnesium has been added in the simulated
seawater, so magnesium served as the second tracer.
Three pore volumes were injected during each diffusion
test in order to guarantee the total displacement. Inert tracers
(barium and magnesium) outlet concentrations were measured.
Steady state tests. The objective of the quasi steady
state tests is determination of dependency “reaction rate
coefficient versus velocity”.
Simultaneous injection of sea and formation waters has
been carried out in steady state tests. The barium outlet
concentration was measured. The outlet concentration tended
to the steady state value during the injection.
Ten pore volumes have been injected during each steady
state test (3.1, 3.2) in order to guarantee the steady state flow
and reaction conditions.
The effluent steady state barium concentration was used
for the data treatment and calculation of dependency Ka(U).
Therefore, the steady state tests have been performed for
different flow velocities.
Transient precipitation tests. Transient
precipitation tests were performed in order to verify the
unsteady state model (2) in general, and specifically the
dependency “reaction rate coefficient versus velocity” as
obtained from the quasi steady state tests.
The displacement of formation water by the seawater has
been performed until the total sweep of the core by the
3
injected water. Afterwards, the core have been washed up and
filled by the formation water again. Second seawater injection
has been carried out at different velocity. Barium
concentration at the core outlet was measured on-line. During
each transient test, three pore volumes have been injected
(2.2, 2.3).
The verification of the model (2) has been performed by
comparison between the laboratory test data and the
simulation results.
Results of Laboratory Data Treatment
In this section are presented the dispersivity values as
obtained from diffusivity tests, the rate coefficient for different
flow velocities as calculated from the quasi steady state tests,
and results of comparison between the laboratory and
modelling data for transient reactive flows.
Dispersion Coefficient from Diffusivity Tests.
Ten alternate injections were carried out at velocities
varying from 7.88x10-7 to 1.58x10-5 m/s.
Dependence of the coefficient of diffusion caused by
dispersion in porous rock, on flow velocity is1,15:
D = αD U (6)
Here αD is a linear size of the core micro heterogeneity. In
this case, the dimensionless diffusion εc is velocity
independent, εc =αD/L, so the diffusion-advection equation (B-
1) in dimensionless co-ordinates (B-3) does not depend on U.
So, the breakthrough curves obtained from different velocity
tests, as plotted versus dimensionless time T, should coincide.
It allows to plot the inlet concentrations as obtained in
different flow velocity tests on the same (T,C) plane. Fig. 4a,b
shows that the “outlet concentration versus dimensionless
time” plots as obtained from tests with different velocities, fit
the same curve.
The transformation c → 1-c allows fitting decline
breakthrough curves (Fig. 4b) to the increasing curves
(Fig. 4a).
The exact solution for advective-diffusive problem
(Appendix B) has been used for determination of diffusion
coefficient D. The results are presented in Fig. 6 for two
different cores that have been used further in the tests with
chemical reactions. The linear dependence D(U)= αDU allows
good adjustment of the data.
For two cores, the αD values found are 0.018m and
0.011m, which is quite reasonable for 0.1 m long cores.
In tests performed, the formation water contained barium
ions, and magnesium ions were presented in the injected
water. So, for the same test, if the outlet barium concentration
decreased, the outlet magnesium concentrations increased, and
vice versa (Fig.5).
The effective (convective) diffusion in porous media (6)
highly exceeds the molecular diffusion15, so the diffusion
coefficients as obtained from flow in porous media tests
should be the same for different components.
The diffusive coefficient values as obtained by equations
(B-3), (B-5) from Ba and Mg breakthrough curves are almost
the same. It validates the results of diffusivity tests. It also
allows recommending magnesium to be used as a tracer.
4 SPE 87457
Reaction Rate Coefficient for Different Velocities
from Steady State Tests The system of governing
equations (2) as applied for steady state flows can be reduced
to a single second order ordinary differential equation (ODE)
(C-13).
The boundary conditions for the second order ODE
correspond to given reagent concentrations at core inlet (C-3)
and to the diffusion absence at the core outlet (A-11). For the
case α<<1 the equation (C-13) subject to boundary conditions
(C-3) and (A-11) allows for the explicit asymptotic solution
(C-15)-(C-17).
The objective of the study was to determine the kinetics
parameter εk from the barium steady state effluent
concentration C(X=1). So, the value C(X=1) as obtained in
laboratory coreflood, was fitted to the analytical solution at the
effluent (C-18) by adjusting the kinetics number εk.
Tables 3,4 presents the steady state outlet concentrations
for two cores as obtained experimentally for different
velocities. It allows calculating the reaction rate coefficient Ka
as a function of flow velocity using the expression (A-6)
for εk.
Fig. 7,8 show the reaction rate coefficient Ka versus
velocity for two cores. A straight line adjusts the experimental
points fairly well:
Ka = λ U (7)
The proportionality coefficient λ is called the kinetics
coefficient. It equals to reciprocal of the average distance that
the ion travels in the rock before being consumed by the
chemical reaction.
Values of λ for two cores are 3003 and 3951 (M m)-1 (Fig.
7a, 8a).
The proportionality between the reaction rate number and
flow velocity (7) takes place for catalytic chemical reactions
for low flow velocities, where the reaction is contro1led by
diffusion17.
Substitution of the expression for reaction rate coefficient
(7) into equation for kinetic number εk (A-6) results in the
system of governing equations in dimensionless co-ordinates
(1). The system (1) is independent of flow velocity. It allows
to treat transient test data on a core outlet C(X=1,T), as
obtained from different velocity tests, on the same plane
(T,C). The outlet concentrations as obtained from different
velocity tests should fit the same breakthrough curve
C(X=1,T).
Fig. 9 shows the effluent barium concentration data as
obtained from two tests on the same core carried out at
different velocities; the points fit the same breakthrough curve
that confirms the above mentioned conclusion.
The dependency of reaction rate coefficient on velocity
shows that the rate should depend on the flow in porous
media. The reaction rate coefficient as obtained from chemical
reactor test is independent of flow dynamics (it is taken at the
so called stirring rate)17-20.
So, Ka data should be obtained from the coreflood data
rather than from the chemical reactors without porous media.
The dependency of reaction rate coefficient on flow
velocity must be used in reservoir simulators5.
Verification of the Model by Transient Tests
Transient tests have been carried out for the second core at
two different velocities, which differ by a factor of three
(3x10-5 m/s and 8.9x10-5 m/s). The reaction rate coefficients
Ka have been evaluated for these two velocities by formula
(7), where the coefficient λ= 3951 (M m)-1 for the second core
(Fig. 8). The dispersivity (heterogeneity scale) for the second
core is αD=0.018 m.
These Ka and αD values were used in the model (2) for
laboratory transient test simulation.
The system (2) for the transient test conditions has been
solved numerically. The details of the finite difference method
used and the Crank-Nicolson linearization applied can be
found in21.
The barium concentrations at the outlet as obtained
numerically and from the transient laboratory tests have been
compared in order to validate the mathematical model (2) and
the data on kinetics coefficient λ as obtained above from the
quasi steady state tests.
As it was mentioned above, the dimensionless system (2)
is independent of flow velocity where Ka has a linear
dependency on velocity, (7). So, outlet barium concentrations
C(X=1,T) were plotted at the same (C,T) plane for two tests
with different velocities.
The results are shown in Fig. 9. The square and triangle
points correspond to outlet concentrations as obtained during
two corefloods with two above-mentioned velocities. The
laboratory data are adjusted by the model (2) with reaction rate
given by the law of mass action.
The experimental points C(X=1,T) from two different
velocity tests are located on the same breakthrough curve. This
observation agrees with the fact that the system (2) is
independent on flow velocity. It validates the
model developed.
Nevertheless, the simulated curves do not agree very well
with experimental data in both cases of the chemical rate
shape. The simulated concentration wave lags behind the
experimentally measured wave in both cases.
Let us explain the difference between modelling and
experimental data.
The difference between equilibrium and non-equilibrium
states during a chemical reaction with precipitation is given by
saturation index. In the barium sulphate case, the saturation
index equals
γ 2 c Ba c SO 4
SI = (8)
K sp
In case of SI=1, precipitation of barium sulphate is
equilibrated by its dissolution in water. Precipitation
dominates for SI>1.
The saturation index (SI) of the aqueous solution of Ba2+
and SO42- at the core outlet as obtained from modelling data is
shown in Fig.9. The SI reaches maximum value 84 for the
case of the mass action law.
The SI value highly exceeds unity. Fig. 9 shows that
SI>>1 during almost all the period of mixture production.
Therefore, barium sulphate precipitates after passing the core,
in the tubes that connect core outlet with the automatic sample
collector. So, in reality the barium concentration at the core
SPE 87457
outlet is almost certainly higher than that measured in the
samples collected during the tests performed.
Propagation of chemical reaction and diffusion
waves.
Let us discuss concentration profiles for inert tracers and
for reagents.
Fig. 10 presents concentration profiles of barium and
sulphate with and without chemical reaction for T=0.25 pvi
(Fig. 10a) and for T=0.50 pvi (Fig. 10b). Dotted lines
correspond to inert tracer concentrations; continuous lines are
plots of the system (2) solution.
Chemical reaction results in decrease of reagent
concentrations if compared with the advection-diffusive tracer
flow. The injected inert tracer arrives to the core outlet with a
small delay comparing with the sulphate arrival.
The difference between concentrations of barium and
displaced inert tracer is higher. This is due to sulphate excess
from seawater over the barium from formation water, at
laboratory tests performed. Fig. 10 shows that decrease in Ba
concentration at the core outlet happens before the inert tracer
concentration decrease. This fact could be used for
interpretation of the produced water chemical analysis data in
scaled-up wells.
The chemical reaction takes place inside the tracer mixture
zone formed by diffusion. Reagent concentrations outside this
zone equal either zero or unity. The chemical reaction rate is
proportional to the concentration product (SI) for the case of
mass action law (A-2). The plot of this product is shown in
Fig. 10 by continuous line for reagents and by dotted line
for tracers.
The position of the product maximum characterises the
centre of chemical reaction zone for reagents, and the centre of
diffusive mixture zone for tracers. Fig. 10 shows that the
diffusion wave centre lags behind the chemical reaction wave
centre. Comparison between the product curves at T=0.25 pvi
and T=0.50 pvi shows that the position difference increases
with time.
The tracer flow could be considered as a particular case of
the reaction-diffusion flow (2) with infinitely slow reaction,
εk→0. The higher is the dimensionless reaction rate coefficient
εk, the higher is the velocity of the reaction wave centre, in
comparison to the diffusive wave centre.
Determination of kinetic coefficient from other
laboratory studies
In the work11, simultaneous injection of formation and sea
waters has been carried out. The sandpack with permeability
k=30 md and porosity φ=0.37 was flooded simultaneously by
seawater with sulphate concentration cSO40=3.1*10-2 M and by
formation water with barium concentration cBa0=1.8*10-3 M.
Barium concentration have been measured at the core effluent.
The barium concentration reached at the steady state cBa
(X=1)=3.5 M after 1 p.v.i.
The dispersivity has not been measured; so we fixed three
values for dispersivity coefficient that are typical for high
permeability homogeneous sandpacks: αD=5*10-3, 10-2 and
3*10-2 m.
5
The explicit expression for the outlet barium concentration
(C-18) permits calculation of kinetic coefficient from
outlet concentration.
The kinetic coefficient values for the above mentioned
values for dispersivity are λ=798, 833 and
963 (M*m)-1 respectively.
The values for kinetic coefficient obtained have the same
order of magnitude as the ones obtained in the
previous section.
Simultaneous injection of barium- and strontium-rich
formation water and seawater containing sulphate ions have
been performed in12. In order to make a rough estimate, we
aggregated barium and strontium in one pseudo component.
The kinetic coefficient varies in the same interval as the one
obtained from the data11.
Comparison of reaction kinetics with and without
porous media
Let us compare the kinetic coefficient values as obtained
above for porous media and for so called batch precipitation
tests as performed in chemical reactors23,24.
Usually the reactor tests are performed at high flow
velocities in so-called kinetics domain, where the reaction rate
is independent of velocity17. The fact of proportionality
between the reaction rate and flow velocity as obtained for
reactive flow in porous media, (7), shows that the reactive
flows under conditions of tests performed lie in the
diffusive domain.
The kinetics constant for diffusive domain λ should exceed
the ratio Ka/U as obtained for the kinetics domain. This fact
could be used in comparison of kinetics constants as obtained
in situ and outside porous media.
Laboratory tests in the current work and in papers23,24 have
been performed using different brines; specific surface areas
were different also. In order to compare a self-consistent data,
let us express the kinetic coefficient as a function of specific
surface area As and of mean ionic activity coefficient γ.
Chemical reaction rate is proportional to the specific surface
area17-20. It is also proportional to product of ionic activities for
Ba2+ and SO42-; ionic activity of component equals mean ionic
activity coefficient times component concentration17-20.
Comparing these speculations with expressions for reaction
rate (A-2),(7), we obtain the following expression for
kinetic coefficient:
λ = λ0 As γ 2 (9)
Here λ0 is the proportionality coefficient.
The results are presented in Table 5. The second and third
columns present data of chemical reactor tests23,24. Diluted
solutions were used in these works, so one can assume that the
solutions are nearly ideal and mean ionic activity coefficient
equals unit. Mean ionic coefficient for highly concentrated
brines, used in the current work, was calculated using Pitzer’s
equations by software OKSCALE25. It equals 0.105.
Velocity of flow in the test23 is not given, so it was taken to
be the same velocity as in reference24.
Values for specific surface areas for reactor tests have been
presented in the corresponding papers23,24. The specific area
was not measured in tests treated in the current work13,14.
6 SPE 87457
Therefore, it was calculated by the formula for specific surface
area for parallel capillary tubes26.
The coefficient λ0 for the reactor test23 is almost three
orders of magnitude higher than for tests24. It could be
explained by the fact that temperature in tests23 was
significantly higher than the one in tests24.
The value for coefficient λ0 as obtained from the tests in
porous media13,14 lies between the two above
mentioned values.
For rigorous study of the porous media effect on rate
coefficient for barium sulphate precipitation, the reactive flow
tests in a core and in reactor should be performed with the
same brines and temperature. The tests should be performed
for the total velocity interval covering both diffusive and
kinetics domains.
Visualization of BaSO4 Crystals by Scanning
Electron Microscopy
The formation of Barium Sulphate crystals during the
injection and formation water mixing in porous media have
been proved by scanning electron microscopy (SEM) –
micrographs show BaSO4 crystal formation in porous space.
Fig. 11, 12 present SEM micrographs of cross-sections
taken at the middle of the core and near to inlet, respectively
(6.5 cm and 0.7 cm counting from the core inlet).
Pores and other pore space elements are presented in Fig.
11a by black patterns, grains are given in clearer colour, and
white crystals are barium sulphate. Fig. 11b shows zoom of
the pattern highlighted in Fig. 11a. The barium sulphate
crystal reference size is about 1-2 µm. The X-ray analysis
identified precipitated crystals as barium sulphate.
The amount of precipitant in the inlet cross section is
higher than the precipitant amount in the middle of the core
(Fig. 12a,b). The crystal size near to the inlet is also
higher – 5-6 µm.
Grains could be seen very clear in Fig. 12a. In zoomed
image (Fig. 12b) one could see the curvilinear grain surface.
The precipitated crystals accompany the grain surface, so they
are stuck to the grain surface.
The image analysis shows that barium sulphate
precipitation takes place in the overall core and not in the
inlet only.
Conclusions
Coreflood tests with injected and formation waters and the
data treatment by an analytical reactive flow model allow
making the following conclusions:
1. A new coreflood sequence/methodology for
laboratory determination of kinetic rate coefficient includes
three test series: diffusion tests with a tracer, transient tests
with chemical reaction and steady state tests with chemical
reaction. Tests are to be performed at several flow velocities.
2. The reaction rate coefficient for Barium Sulphate
scaling is proportional to flow velocity within the range of
studied values (two orders of magnitude variation).
This fact must be used in scaling modelling around
production wells where flow velocity varies two orders of
magnitude from the wellbore up to the reservoir.
3. The dimensionless system of governing equations is
independent of flow velocity. So, the component breakthrough
curves as obtained in different velocity tests are to be fitted by
the same curve, and it was observed in transient tests.
4. The relatively poor agreement between the modelling
and transient test data is explained by possible precipitation in
the setup tubing that has not been controlled during the tests.
5. The chemical reaction wave propagates in porous
media faster than the centre of the diffusive mixture zone.
Acknowledgement
Authors thank Dr. Antonio Luis S. de Souza, Alexandre G.
Sequeira, Dr. Farid Shecaira and Maylton Silva (Petrobras),
Prof. Themis Carageorgos (UENF-Lenep), Dr. Oleg Dinariev,
Prof. A. D. Polianin and Prof. V. V. Dilman (Russian
Academy of Sciences) for fruitful discussions.
Many thanks are due to Prof. Dan Marchesin (IMPA) for
the development of the numerical model for scaling.
Authors thank Petrobras for permission to publish the
paper. Many thanks are due to technical stuff of Cenpes
laboratory – Marcia, Rosilene, Nelia and Rodrigo.
Nomenclature
c - molar concentration in aqueous solution, M
C - dimensionless Ba concentration
Y - dimensionless SO4 concentration
D – diffusion/dispersion coefficient, m/s2
k - initial permeability, D
β - formation damage coefficient
Ka - reaction rate coefficient, (M s)-1 (2nd order reaction)
Ksp - solubility product number, M2
L - core length, m
M - molar unit for concentration equals gmol/L
(equals kgmol/m3)
p - pressure, Pa
q - chemical reaction rate, M/s
t - time, s
T - dimensionless time
U - flow velocity, m/s
x - linear co-ordinate, m
X - dimensionless linear co-ordinate
Greek letters
α - ratio between the initial concentrations of Ba and SO4
in formation and injected waters
αD – dispersivity coefficient, m
εD - dimensionless diffusive number
εk - dimensionless chemical kinetics number
φ - porosity
γ - mean ionic activity coefficient
λ - kinetic coefficient, (M m)-1
µ - viscosity, Pa s
σ - molar concentration of solid deposit, M
Subscripts
Ba, SO4 - ions
Superscript
0 - concentrations in injected fluid
SPE 87457 7

References Petroleum Engineering Conference held in Port-of-Spain,

1. Sorbie, K. S. and Mackay, E. J., 2000, Mixing of Injected, Trinidad, West Indies, 27–30 April
Connate and Aquifer Brines in Waterflooding and its 17. Fogler, S., 1998, Chemical Reactions Engineering,
Relevance to Oilfield Scaling, Journ. Petr. Sci. Eng., 27, Prentice Hall, NY.
85-106. 18. Stumm, W., 1992, Chemistry of Solid-Water Interface,
2. Rosario, F. F., Bezerra, M. C., 2001, Scale Potential of a John Wiley and Sons, NY, 428 p.
Deep Water Field – Water Characterisation and Scaling 19. Bethke, C., 1996, Geochemical Reaction Modelling,
Assessment, SPE paper 68332 presented at 2001 SPE Third Oxford University Press, 397 p.
International Symposium on Oilfield Scaling, Aberdeen, 20. Nielsen, A. E., 1959, The Kinetics of Crystal Growth in
UK, 30-31 January. Barium Sulphate Precipitation II – Temperature
3. Oddo, J.E. and Tomson, M.B., 1994, Why Scale Forms and Dependence and Mechanism. Acta Chemica Scandinavica
How to Predict It, Journ. SPE PF, February, p.47-54. 13 (784-802).
4. Rocha, A., Frydman, M., Fontoura, S., Rosario, F., Bezerra, 21. Beam, R. M., and Warming, R. F., 1976, An implicit finite
M., 2001, Numerical Modelling of Salt Precipitation during difference algorithm for hyperbolic systems in
Produced Water Reinjection, SPE 68336 presented at 2001 conservation-law form, J. Comput. Phys., 22, 87-116.
SPE Third International Symposium on Oilfield Scale, 22. Bezerra, M. C., Rosario, F. F., Khalil, C. N., Milanez, P.V.,
Aberdeen, UK, 30-31 January. 1996, Prevention and Control in Namorado Field, Campos
5. Delshad M., Pope, G. A., 2003, Effect of Dispersion on Basin, Brazil presented 1996 NACE International Annual
Transport and Precipitation of Barium and Suphate in Oil Conference and Exposition.
Reservoir, SPE paper 80253 presented at the SPE 23. Christy, A. G., Putnis, A., The Kinetics of Barite
International Symposium on Oilfield Chemistry held in Dissolution and Precipitation in Water and Sodium Chlorite
Houston, Texas, USA, 5-7 February. Brines at 44-850C, Geochimica et Cosmochimica Acta,
6. Woods, A.W., Parker, G., 2003, Barium Sulphate 1992, 57, 2161-2168.
Precipitation in Porous Rock Through Dispersive Mixing, 24. Nancollas, G., Liu, T., 1975, Crystal Growth and
SPE paper 80401 presented at SPE 5th International Dissolution of Barium Sulphate, SPE paper 5300 presented
Symposium on Oilfield Scale, Aberdeen, UK, 29-30 at 1975 SPE/AIME Oilfield Chemistry Symposium, Dallas,
January January.
7. Araque-Martinez, A., Lake, L. W., 1999, A Simplified 25. Atkinson, G., Raju, K., The Thermodynamics of Scale
Approach to Geochemical Modelling and its Effect on Precipitation, SPE paper 21021 presented at 1991 SPE
Well Impairment, SPE paper 56678 presented at the 1999 Simposium on Oilfield Chemistry held in Anaheim, Ca, 21-
SPE Annual Technical Conference and Exhibition held in 23 February.
Houston, TX, 3-6 October 26. Amix, B., Bass, R. and Whiting, A.: “ Applied Reservoir
8. Philips, O.M., 1991, Flow and Reactions in Porous Media, Engineering, McGraw Hill Book Co, NY (1964), 171-174.
Cambridge Univ Press.
9. Bedrikovetsky, P. G., 1994, Mathematical Theory of Oil Appendix A. Governing Equations
and Gas Recovery, Kluwer Academic Publishers, The mass balance for ions Ba2+, SO42- and for BaSO4
London/Boston molecules is6-8,14:
10. Green, D. W. and Willhite, G. P., 1998, Enhanced Oil
⎧ ∂c Ba
⎪φ ∂t + U ∇c Ba = ∇ ( D ∇c Ba ) − q
Recovery, SPE Textbook Series.
11. Wat, R.M.S., Sorbie, K.S., Todd, A.C., 1992, Kinetics of
BaSO4 Crystal Growth and Effect in Formation Damage, ⎪ ∂c
⎪
⎨φ + U ∇c SO 4 = ∇ ( D ∇c SO 4 ) − q
SO 4
SPE paper 23814 presented at SPE Intl Symposium on (A-1)
Formation Damage Control held in Lafayette, Louisiana,
⎪ ∂t
⎪φ ∂σ = q
February 26-27.
12. Todd, A.C. and Yuan, M.D., 1992, Barium and Strontium
Sulphate Solid-Solution Scale Formation at Elevated ⎪⎩ ∂t
Temperatures, SPE Production Engineering, February, 85- The law of mass action is assumed for the chemical
92.
reaction (salt deposition) rate17
13. Lopes Jr., R. P., 2002, Barium Sulphate Kinetics of
Precipitation in Porous Media: Mathematical and q = K a c Ba c SO 4 (A-2)
Laboratory Modelling, in Portuguese, MSc Thesis, North
The modified Darcy’s law includes the permeability
Fluminense State University-Lenep/UENF, Macaé, RJ,
Brazil.
damage due to salt precipitation:
14. Bedrikovetsky, P. G., Lopes Jr., R.P., Rosário, F. F, k
U=− ∇p (A-3)
Bezerra, M. C. and Lima, E. A., 2003 Oilfield Scaling –
Part I: Mathematical and Laboratory Modelling, SPE
µ (1 + β σ )
paper 81127 presented at the SPE Latin American and The system of five equations (A-1)-(A-3) is closed. The
Caribbean Petroleum Engineering Conference held in Port- unknowns are three concentrations cBa, cSO4, cBaSO4 , pressure p
of-Spain, Trinidad, West Indies, 27–30 April . and flux U.
15. Nikolaevskii, V. N., 1990, Mechanics of Porous and The linear problem describes one-dimensional flow during
Fractured Media, World Scientific Publishing Co., laboratory coreflooding:
Singapure.
16. Bedrikovetsky, P.G., Gladstone, P.M., Lopes Jr., R. P.,
Rosário, F. F., and Bezerra, M. C., 2003, Oilfield Scaling –
Part II: Productivity Index Theory, SPE paper 81128
presented at the SPE Latin American and Caribbean
8 SPE 87457

∂c Ba ∂c Ba ∂ 2 c Ba If the mixture zone is several times less than the core

φ +U =D − K a c Ba c SO 4 length, the process can be described by the following
∂t ∂x ∂x 2 initial conditions:
∂c ∂c ∂ 2 c SO 4 ⎧c 0 , x ≤ 0
φ SO 4 + U SO 4 = D − K a c Ba c SO 4 (A-4) t=0: c=⎨ (B-2)
∂t ∂x ∂x 2
∂σ ⎩0 ,x>0
φ = K a c Ba c SO 4 The solution of the problem (B-1), (B-2) is15:
∂t
⎛ ⎞
k ∂p ⎜ x −U t ⎟
U=− c0 φ
µ (1 + β σ ) ∂ x c( x , t ) = erfc⎜ ⎟ (B-3)
2 ⎜ Ut ⎟
It is assumed that the diffusion coefficients for Ba and SO4 ⎜ 2 αD φ ⎟
ions are equal and proportional to flow velocity: ⎝ ⎠
DBa ≅ DSO4 ≅ D = α D U (A-5) The solution for the displacement of traced water by the
sweet water
Let us introduce the following dimensionless parameters:
⎧0 , x ≤ 0
c c SO4 x t =0: c = ⎨ (B-4)
C = Ba
0
Y= X= ⎩1 , x > 0
c Ba c 0SO4 L
Ut c0 σ is obtained from solution (B-3) as c0-c(x,t):
T= α = 0Ba S= (A-6)
φL c SO 4 0
c Ba ⎛ ⎞
⎜ x −U t ⎟
D αD ka L c 0
c 0
φ
εD = = εk =
SO 4 c( X ,T ) = c − erfc⎜
0 ⎟ (B-5)
2 ⎜ Ut ⎟
LU L U ⎜ 2 αD φ ⎟
First two equations (A-3) take the form: ⎝ ⎠
⎧ ∂C ∂C ∂ 2C
⎪⎪ + = εD − εk C Y Appendix C. Steady State Linear Flow
∂T ∂X ∂X 2 Let us consider steady state linear flow in a core:
⎨ (A-7)
⎪ ∂Y + ∂Y = ε D ∂ 2Y
−εk α C Y ∂C ∂Y
= =0 (C-1)
⎪⎩ ∂T ∂X ∂X 2 ∂T ∂T
The displacement of water with Ba cations by SO4-rich Substituting (C-1) into the first two equations (A-7), obtain
water is described by the following initial conditions: the following ordinary differential equations:
T = 0 : C = 1 ;Y = 0 (A-8) ⎧dC d 2C
The injection of water with SO4 anions into the reservoir, ⎪ = ε D − εk CY
saturated by Ba-rich water, corresponds to the inlet boundary ⎪d X d X2
⎨ (C-2)
conditions where fluxes are fixed for both species15:
⎪ 1 dY = εD d Y −ε CY
2

∂C ∂Y ⎪⎩ α d X α d X 2 k
X = 0 :C − ε D = 0 ,Y − ε D =1 (A-9)
∂X ∂X Advection-diffusion fluxes of both components are fixed at
Neglecting the diffusive term simplifies the inlet boundary the core inlet15:
condition (A-9): dC
X = 0 : C = 0 ,Y = 1 (A-10) X = 0 :C − ε D =1 (C-3)
dX
The assumption that an ion does not diffuse back into the
core after leaving the outlet together with the carrier water, dY
X = 0 :Y − ε D =1 (C-4)
results in the Brenner’s boundary condition15: dX
dC dY As a consequence of diffusion neglecting on the core inlet,
X = 1: = =0 (A-11)
the injected concentrations for both reagents are fixed at X=0:
dX dX
X = 0:C = Y = 1 (C-5)
Appendix B. Advection-Diffusion Wave in Linear If the particle already left the core, it would not diffuse
Coreflooding back. The corresponding outlet boundary conditions are given
The advection-diffusion equation describes displacement by (A-11)
of water by traced water in a core: So, the steady state chemicals distribution along the core
∂c ∂c ∂ 2c during the flow is described by a boundary problem (C-3),(C-
φ +U =D 2 (B-1) 4) for the system of two ordinary differential equations (C-2).
∂t ∂x ∂x Let us introduce the following linear combination of
two concentrations:
SPE 87457 9

V( X) = C( X) −
Y( X )
C (1) =
(Γ1 − Γ2 )e Γ 2

( ) (C-18)
(C-6)
α Γ1 − Γ2e (Γ2 − Γ1 ) + ε D Γ2Γ1 e (Γ2 − Γ1 ) − 1
The subtraction of the second equation (C-2) from the first
It is interesting to mention that application of both inlet
one results in the following equation for the function V(X): boundary conditions (C-3), (C-4) and (C-5) results in the same
dV d 2V solution (C-8) for V=V(X). So, the equation for barium
= εD (C-8) concentration C(X) (C-10) is the same for both cases.
dX d X2
Applications of inlet boundary conditions (C-5) also results in
The following inlet boundary condition for V(X) follows explicit solution that differs from (C-15)-(C-17). Nevertheless,
from (C-3): the calculations show that the difference in distributions of
X = 0 : V = 1 − α −1 (C-9) C(X) and Y(X) for both types of inlet boundary conditions is
The outlet boundary condition for V(X) follows from (A- negligibly small for the range of parameters εD and εk in
11): oil reservoirs.
dV
X = 1: =0 (C-10)
dX
Integrating both parts of (C-8) accounting for boundary
conditions (C-9) and (C-10) results in the solution:
V ( X ) = 1 − α −1 (C-11)
So, the concentration difference (C-6) is constant along the
core during the steady state flow.
Expressing the sulphate concentration Y(X) from (C-11)
Y( X) = 1 + α(C( X) − 1) (C-12)
and substituting it into the first equation (C-2), we obtain an
ordinary differential equation for C(X):
d 2C d C
εD = + ε k C ( 1 + α ( C − 1 )) (C-13)
d X2 d X
If sulphate concentration in seawater highly exceeds
barium concentration in formation water, parameter α is
negligibly small, α <<1, (A-6). The term C(X)-1 varies from
minus unity to zero, so the second term in brackets in right
hand side of (C-13) can be neglected comparing with unity.
Equation (C-13) becomes a linear second order ordinary
differential equation
d 2C d C
εD = + εk C (C-14)
d X2 d X
The boundary problem (C-3),(A-11) for equation (C-14)
allows for exact solution:
C ( X ) = c1 e Γ1 X + c2 e Γ2 X
1 1+ 4ε D ε k
Γ1 = + (C-15)
2ε D 4ε D
2

1 1+ 4ε c ε k
Γ2 = −
2ε D 4ε D
2

Two constants in (C-15) are found from boundary conditions

(C-3) and (A-11):
− Γ2e (Γ2 − Γ1 )
c1 = (C-16)
Γ1 − Γ2 e (Γ2 − Γ1 ) + ε D Γ2Γ1e (Γ2 − Γ1 ) − ε D Γ2Γ1

Γ1
c2 = (Γ2 − Γ1 ) (C-17)
Γ1 − Γ2 e + ε D Γ2Γ1e (Γ2 − Γ1 ) − ε D Γ2Γ1
The outlet concentration is calculated by (C-15),(C-16)
10 SPE 87457

Table 2 - Petrophysical properties of samples

Table 1 - Synthetic brine compositions
Con centrations (mg/ L) Sample 1 2

So lution Ba2+ Na+ Mg 2+ SO42- Cl- k(mD) 1 74 667

F ormatio n
229 3 824 9 - - 59098
φ(%) 20.08 21.16
W ater
L(cm) 10.29 12.81
Sea W a ter
- 1 516 9 - 2 834 21299
(p recip it ation) d(cm) 3.67 3.78
Sea W a ter
- 1 150 1 1391 - 21790 pvi (mL) 21.86 29.58
(diffusion)

Table 4 – Core outlet

Table 3 – Quasi steady state tests concentrations for sample
Table
outlet 4 - Steady state
concentrations tests 1
for sample Table 5 - Steady state 2tests
at
outlet concentrations for sample 1 quasi steady state tests
outlet concentrations for sample 2
Sample 1 Sample 2
Seq. U(m/s) PVI CBa (mg/L) Seq. U(m/s) PVI CBa (mg/L)
2.82 9.36 2.09 0.30
1.58E-5

2.97E-6
3.18 7.19 4.93 0.29
1.1

2.1
3.37 9.62 5.87 0.29
CBa med 8.72 CBa med 0.29
1.77 8.72 1.95 2.08
3.15E-5

2.87 8.69 2.42 1.58

1.2

3.96 7.85 2.97E-5

3.83 1.64
2.2

CBa med 8.42 4.76 2.34

1.23 5.81 5.70 1.81
1.58E-6

1.92 5.42 CBa med 1.89

1.3

2.60 5.52 2.49 3.05

8.91E-5

CBa med 5.58 2.96 1.94

2.3

3.90 2.41
CBa med 2.47
2.29 0.86
1.49E-4

2.76 0.64
2.4

Table 5. Comparison between reaction rates 4.17 0.63

in porous media an in reactor 5.11 0.77
Christy e Putnis Nancollas e Liu Sample 2 CBa med 0.72
(1992) (1975) seq 2.5 1.89 0.38
-1
Ka (M s) 7,80 E+03 6,38 E+00 6,40 E-01 2.22 0.40
1.49E-6

2.55 0.45
2.5

U (m/s) 4,70 E-01 4,70 E-01 1,49 E-04 3.21 0.38

-1
3.88 0.40
A (m ) 4,82 E+04 1,38 E+02 5,27 E+06
5.20 0.50
CBa med 0.42
γ 1 1 0,105

λ0 (M )
-1
3,45 E-01 2,82 E-04 7,39 E-02
SPE 87457 11

Mixture
zone
FW Confinement pressure
Ba2+
2+
Ba manometer

SO42- Pressure
BaSO4 Transducers

Diffusor face-B
Diffusor face-A
CORE
Automatic
collector
Figure 1 - BaSO4 precipitation schema
Core Holder
SW
SO42-
Figure 2 - Experimental setup schema

Basic Experimental Setup - Inlet valves, core holder and collector

Pump Diffuser - Face B Diffuser - Face A

Figure 3 - Experimental Setup - Photos

C1
1
Cexp C
0. 5 C(Ba)exp
Cadj
C(Mg)exp
0 0.5
0 0. 5 1 1. 5 2
PVI C(Ba)adjust
(a )
C1 C(Mg)adjust

Cexp 0
0 0.5 1 1. 5 2
0. 5 PVI
Cadj Figure 5- Barium and magnesium concentrations
at the core outlet being adjusted by the same
0
0 0. 5 1 1. 5 2
diffusion coefficient, U =1,5E-5m/s
( b) PVI
Figure 4 - Dimensionless concentrations for Ba2+ and
Mg2+ in increasing (a) and decreasing (b) breakthrough
profiles
12 SPE 87457

ln (D) x ln (U)
-14

-15
Sample 1
-16 ln(D) = 1.0859 ln(U) - 3.8995
R2 = 0.875
ln (D)

-17

-18 Sample 2

-19 ln(D) = 0.8689 ln(U) - 6.3264

R2 = 0.9933
-20
-15 -13 -11 -9
ln (U)
Figure 6 6-–Diffusion
Figure coefficient
Diffusion coefficient versus
versus velocity velocity for two cores
for two cores

Ka U
Ka U

Figure 7 – Rate constant – velocity dependence Figure 8 – Rate constant versus flow velocity
for sample 1 for sample 2

1,0 90
C,Y SI
0,8

60
0,6
C
Y
0,4 U=3.0E-05
30
U=8.9E-05
0,2 SI

0,0 0
0,0 0,2 0,4 0,6 0,8 1,0
T

Figure 9 – Breakthrough curves during transient test by laboratory

and mathematical modelling
SPE 87457 13

1,0
150
C,Y
SI
C
0,8
Y

100
0,6 FW(fraction)

(a) SW(fraction)

0,4
(FWxSW) /
50 (FWxSWmax)
SI
0,2

0,0 0
0,0 0,2 0,4 0,6 0,8 1,0
X

1,0
SI
C,Y 90
C
0,8
Y

0,6 FW(fraction)
60

SW(fraction)
(b)
0,4
(FWxSW) /
(FWxSWmax)
30
SI
0,2

0,0 0
0,0 0,2 0,4 0,6 0,8 1,0
X

Figure 10 - Concentration and saturation index profiles along the core;

(a) T=0,25 PVI; (b) T=0,50 PVI.
14 SPE 87457

(a) (b)
Figure 11 - SEM micrographs of the porous rock taken from the
middle of the core: (a) image on the grain scale;
(b) zoomed image

(a) (b)

Figure 12 - SEM micrographs of the porous rock taken near

to the core inlet: (a) image on the grain scale; (b) zoomed image