A Study of Solubility of Strontium Sulfate
Donald F. Jacques, SPE, Exxon Chemical Co.
Brent I. Bourland, Exxon Chemical Co.
Abstract
This paper describes the results of a solubility study of
strontium sulfate in sodium chloride brine. A predictive
equation for the solubility of strontium sulfate is
presented, which is useful for calculating solubility in
water containing 0 to 200 giL (0 to 200 g/dm 3 ) sodium
chloride at temperatures from 100 to 300°F (38 to
149°C), with pressures from 100 to 3,000 psig (689 to
20 684 kPa), and total ionic strengths of 0 to 3.43. This
paper also describes the experimental technique
employed and the special equipment designed for this
study.
Introduction
Strontium sulfate scale formation has become a growing
concern in oil-production systems. 1-3 Until recently, the
appearance of strontium in oilfield scales has been
primarily in the presence of barium sulfate scale. Almost
pure SrSO 4 scale now is observed in several production
wells around the world. The scale formation is primarily
a result of subsurface commingling of waters, which
results in a water supersaturated in SrSO 4.
The literature does not readily provide solubility data
that permit prediction of SrSO 4 scaling under downhole
conditions. Jacques et al. 4 presented the most recent
SrSO 4 solubility data in 1979. Their work presented a
comprehensive literature survey, new solubility data
developed in synthetic brine systems from 77 to 212°F
(25 to 100°C), and studies of effective SrS04 scale in-
hibitors. The study showed that the solubility of SrS04
increased with increasing ionic strength and decreased
with increasing temperature once 104 ° F (40 °C) had
been exceeded. A method of predicting SrSO 4 precipita-
tion in field brines was suggested on the assumption that
0197·752018310004·9625$00.25
Copyright 1983 Society of Petroleum Engineers of AIME
292
solubilities based on pure sodium chloride systems were
the limiting thermodynamic case. This model did not
provide for pressure effects and it was limited to 212°F
(l00°C). Some work 5 ,6 available at the time indicated
slight increases in SrS04 solubility with pressure, but
the study was limited to 95°F (35°C) maximum and low
salinity.
To predict SrS04 precipitation and scaling under
downhole conditions, solubility data are needed that
bracket all possible temperatures, pressures, and
salinities of oilfield waters. This study reports the results
of new solubility determinations developed under
broader test conditions. The results then are combined
with the 1979 data 4 to provide a predictive solubility
model.
Theory
To develop data that could be used generally to predict
SrS04 solubility in most oilfield brines, the solubility of
SrS04 was studied in pure synthetic NaC! brines. The
salinity, temperature, and pressure ranges used in this
study were selected to encompass most oilfield condi-
tions.
Minimum Maximum
Salinity, giL, NaCl o 200
Pressure, psig (kPa) 100 (689) 3,000 (20 7(0)
Temperature, OF (0C) 100 (38) 300 (149)
To develop the most useful data possible over the ranges
of the three variables listed demanding the fewest possi-
ble experiments, we used the Box-Behnken experimental
design as taught by DuPont. 7 This space-filling design
was followed by a second design, a 2 3 factorial, inside
the first design.
The Box-Behnken approach for three independent
variables simply defines a cube with the three variables
SOCIETY OF PETROLEUM ENGINEERS JOURNAL
 TABLE 1-BOX-BEHNKEN DESIGN VARIABLE LIMITS
Low Midpoint High
Temperature, of 100 185 300
Temperature, (1/K)x 10 3 3.216 2.793 2.369
Salinity, 9 NaCI/L 0 52.25 200
Salinity, ~ 0.053 0.955 1.858
Pressure, psig 100 1,550 3,000
represented by the x, y, and z axes. The limits of each
variable selected become the limits of the cube. The
trials to be run experimentally represent the centers of
each edge on the cube and the center point of the cube
(Fig. 1). The general procedure calls for the response,
the SrSO 4 solubility, to be defined in terms of the three
independent variables. It is found generally that the
variables should be used in a form that will yield the
most linear relationship possible with the response. It
should be apparent from thermodynamics that the
solubility of a salt defined as log K sp' log solubility
product constant, is linearly related to liT and the square
root of the ionic strength. It was also anticipated 5,6 that
pressure would provide a linear response with log K sp'
For these reasons, the variables were defined as liT,
.;;;., and p, and the dependent variable (response) was
defined as log K sp' The centers of the edges of the cube
then are defined as the midpoint of the variables as
linearly defined. For example, the midpoint of 100 to
300°F (38 to 149°C) would be 200°F (93°C) if
temperature is defined as OF (0C) but would be 185°F
(85°C) if temperature is defined as reciprocal degrees
Kelvin. Table 1 shows the range of each variable and
midpoint as used in the initial Box-Behnken design for
this study. Each trial (set of variables) was run at least
twice. Regression analyses were run on the data to
develop an equation defining solubility (log K,p) in
terms of liT, p, and';;;'. The regression results were
then used to predict solubility of SrSO 4 for trials within
the Box-Behnken cube (i.e., Trials 13 through 21 in
Fig. I).
Trials 13 through 21 are a typical 2 3 factorial design
using Trial 13 as a common midpoint with the Box-
Behnken design. As shown in Fig. 1, the factorial design
for three variables is also defined by a cube, and the trial
points are at the comers of the cube. In keeping with the
linear definition of the variables, the range of each
variable in the factorial design is shown in Table 2.
The factorial trials were determined experimentally
and were compared to the predicted values from the Box-
Behnken regression. A combined regression of the Box-
Behnken and factorial trials were then run to yield an im-
proved predictive equation. We combined the solubility
data from the previous work 4 that fall within the design
of the present study [i.e., the 104 and 150°F (40 and
66°C), 0 to 200 giL (0 to 200 g/dm 3 ) NaCl] with this
study to yield the final predictive solubility model.
Equipment Design
The basic design required that laboratory-prepared
SrS04 crystals be dissolved to an equilibrium condition
in a test brine at temperatures up to 300°F (l49°C) and
at pressures up to 3,000 psig (20 700 kPa). Once
APRIL 1983
TABLE 2-2 3 FACTORIAL DESIGN VARIABLE LIMITS
Low Midpoint High
--
Temperature, of 140 185 238
Temperature, (11K) x 10 3 3.005 2.793 2.581
Salinity, 9 NaCI/L 13.9 52.25 114
Salinity, ~ 0.504 0.955 1.404
Pressure, psig 825 1,550 2,275
equilibrium was obtained, the solution was to be filtered
under conditions as close as possible to those of the test.
The amount of filtrate would be measured and then
brought back to ambient conditions without permitting
precipitation of SrS04. The filtrate then would be
analyzed for strontium by atomic absorption and the ac-
tual SrS04 solubility would be calculated.
A two-pressure vessel system was devised after
several design considerations. SrS04 was dissolved in
the test brine to the saturation point in one vessel, and
then the solution was filtered under the temperature and
pressure conditions of the test into a second vessel that
contained a preweighed amount of fluid. The amount of
filtrate received into the second vessel was determined
by weight after the filtration, and the appropriate
analytical procedures then were used to determine the
SrS04 solubility in the first vessel. The first vessel was
called the "test pressure vessel" (TPY) and the second
vessel was called the "receiver pressure vessel" (RPY).
The TPY and RPY were originally manufactured from
316 stainless steel but were changed to Hastelloy C
because of corrosion problems. Initial studies were per-
formed in air, but it was found that an oxygen-free at-
mosphere was required to prevent Hastelloy C corrosion
in tests above 250°F (121°C); as a result, all tests were
performed under nitrogen atmosphere. Fig. 2 shows the
design of the TPY, which was a symmetric vessel that
could be rolled on its side. Fig. 3 shows the design of the
RPY. It was designed to weigh under 2,500 g to permit
accurate measurements of the amount of filtrate entering
the vessel. Both pressure vessels were fitted with safety
heads set at 4,000 psig (27 579 kPa).
Solubility Test Procedure
The solubility test method consisted of preparing pure
NaCI brines from ACS reagent-grade chemicals and
water that was first deionized and then distilled (DOW).
The brines were filtered through a 0.2-micron (0.2-fJ.m)
membrane and then were deaerated with nitrogen to
remove oxygen. SrS04 crystals were prepared by the
method previously outlined. 4
While under a nitrogen atmosphere, 50 mL (50 cm 3 )
of deaerated test brine and 0.25 g of laboratory-prepared
SrSO 4 were added to the TPY. The top was secured on
the TPY, which then was pressurized with nitrogen at
room temperature. The pressure selected was calculated
using the ideal gas law so that when the test temperature
was achieved, the test pressure would also be achieved.
The loaded, pressurized TPY then was checked for leaks
and placed in a modified roller oven (Baroid Model 701)
for 72 hours. The roller oven could hold four TPY's.
Rolling was used to provide constant agitation of the test
fluids. The oven temperature-control system was
293
 modified by replacing the temperature controller with an
5 RFL Model 720 zero-voltage-firing proportional
temperature controller. This permitted the oven
7
: 19~ ________-=~
temperature to be controlled to within ±0.2°F
,, (±O.loq.
Before the end of the dissolution phase of the study in
2. 18:I

I ,
1 the TPV, the RPV was charged with a known weight of
test brine or DDW and pressurized. The desired RPV

4
,
I
I
I
I
,
I
I
I
I
•
13 working pressure was selected using the ideal gas law, so
that it would permit filtration of a suitable amount of test
:: .--------
, / 15 16
solution at a minimal pressure differential between the
TPV and RPV. The weight of the RPV was determined
l-j~-------~---- both before and after charging using a Mettler PC 4400
digital top-loading balance. Weights were measured to
12,," 6
±0.1 g.
,
" 8

Trials 1 - 13: 1.t Design: Box-Behnken

After 72 hours in the roller oven, the TPV was placed
upright in a Precision Model 28 forced-draft oven. The
oven was set at the same temperature as the roller oven.
....- Trials 13 - 21: 2nd Design: 2' Factorial
A charged RPV and a filtration assembly were
1/T
equilibrated in the Precision oven before the placement
Fig. 1-Experimental strategy. of the TPV. The filtration assembly (Fig. 4) consisted of
a high-pressure, stainless steel, 25-mm (0.98-in.)
millipore filter, XX45 025 00, connected to an
Autoclave series SW vent-type valve (labled "C" in
Fig. 4). All connecting tubing and fittings were ;Is-in.
(3.048-mm) 316 stainless steel. The filter membrane was
a Millipore® Fluoropore filter, unlaminated, 0.5-micron
(0.5-Jlm) pore size, FHUP 025 00. The TPV, filtration
assembly, and RPV were connected in the Precision
oven at Points Wand X. A ;Is-in. (3.048-mm) nitrogen
line was connected at Points Y and Z, as shown, inside
the oven. Note that Valves A, B, and D are closed during
100 !ilL VESSEL this step. This line went through the oven wall and to a
valve-and-gauge assembly that was connected to a high-
Fig. 2-Test pressure vessel (TPV) design. pressure nitrogen source. The high-pressure source con-
sisted of a Teledyne/Sprague air-operated gas booster
(Model S-486-CN -100), which took 400-psig
(2800-kPa) nitrogen from a cylinder and compressed it to
5,000 psig (34500 kPa) into a 300-mL (300-cm 3 )
Autoclave pressure vessel (Model BC-030-5A). The
regulators used were Grove 15LHX and 15LX small-
volume, high-pressure units that could operate up to 650
and 3,000 psig (4500 and 20 700 kPa) , respectively. The
I • ,
~

pressure gauges [100, 1,000, and 3,000 psig (689, 6895,

7S !ilL VESSEL and 20 700 kPa)] used were Marsh Master Test gauges
(ANSI B40.1, Grade 3A) with an accuracy of ±0.25%
Fig. 3-Receiver pressure vessel (RPV) design.
of span.
Once the system was assembled (Fig. 4), the oven
door was closed and the system brought back to thermal
equilibrium. Valve C then was opened. (Note that Valve
C opens and closes only Connection Z and that Connec-
High PreMur. tions Wand X are always open.) Valves E and F were
NItrogan
Soun:a then opened and the pressure gauge was brought to the
Inlkla 0utaIda desired working pressure of the RPV. When this was ac-
complished, Valve F was closed to isolate the system.
Oven
WaH The system was then checked for leaks by observing the
pressure gauge for a period of time. If no pressure drop
was observed, Valve D on the RPV was opened. If no
I
Millipole APV leaks had occurred in the RPV during equilibration in the
Filter oven and if the initial pressurization was done properly at.
room temperature, no change was observed on the gauge
Fig. 4-Filtration system. upon opening Valve D. If a slight leak had occurred, the
gauge pressure would drop when Valve D was opened.
The actual pressure in the RPV could be calculated since
294 SOCIETY OF PETROLEUM ENGINEERS JOURNAL
 TABLE 3-SrSO 4 SOLUBILITY DATA-BOX-BEHNKEN DESIGN

SrS0 4
Experimental Predicted
p Temperature NaCI Salinity SrS0 4 (mg/L)
Trial Run (psig) (OF) (gIL) (~) (mg/L) Eq. 1 Eq. 2
--
65202 3,135 301 52.25 0.947 411 388 448
65226 3,040 305 52.25 0.947 407 383 441
65227 3,095 305 52.25 0.947 379 384 443

2 63962 2,932 99 52.25 0.948 398 731 656

65214 2,986 101 52.25 0.948 694 726 655

3 63969 96 303 52.25 0.947 355 342 404

65202 100 301 52.25 0.947 315 343 406

4 65212 96 101 52.25 0.948 587 646 635

65213 95 101 52.25 0.948 599 646 635

5 63975 3,135 187 200 1.853 890 819 820

65222 3,135 187 200 1.853 777 819 820

6 63965 3,068 186 0 0.0461 99 83 92

63974 2,987 187 0 0.0432 87 82 91

7 63967 100 186 200 1.853 665 723 768

65223 99 187 200 1.853 706 724 767

8 63968 100 186 0 0.0520 126 74 86

65221 81 187 0 0.0424 84 73 85

9 65225 1,550 305 200 1.853 799 747 655

65241 1,536 300 200 1.853 777 747 659
65240 1,550 300 200 1.853 646 747 659

10 65207 1,170 301 0 0.0258 31 41 31

65218 1,570 310 0 0.0285 38 40 29
65219 1,587 310 0 0.0266 33 40 29

11 65210 1,482 101 200 1.853 867 787 859

65215 1,400 101 200 1.853 806 783 858

12 63964 1,550 100 0 0.0470 103 146 152

63978 1,536 100 0 0.0540 136 146 153
63208 1,496 101 0 0.0571 152 145 152

13 63966 1,604 186 52.25 0.948 519 500 528

63973 1,604 187 52.25 0.948 574 499 527
65220 1,404 187 52.25 0.948 518 499 527

the volume of the system from Valve F to Valves A, B,
and D had been detennined. If the pressure loss was
significant, the RPV was replaced with another one.
Valve C was then closed.
The TPV working pressure was checked in a manner
similar to that described for the RPV. The gauge
pressure was set at the desired TPV test pressure by ad-
justment of the regulator. Valve F was closed and then
Valve A was opened. The actual pressure in the TPV
was calculated. If a significant leak was detected, the
filtration was not perfonned and the test was discarded.
The filtration was perfonned if the TPV was judged to be
in good condition. The calculated pressure in the TPV
was recorded as the actual test pressure.
The filtration was perfonned by opening Valve F to
provide a pusher pressure close to the test pressure.
Valve B was opened. The filtration was complete in less
than 1 second. Then, Valves A, B, D, and E were
closed. The nitrogen system was vented through Valve G
and disconnected at Points Y and Z. The filtration
APRIL 1983
assembly was then vented through Valve C and discon-
nected from the TPV and RPV at Points Wand Z. A
pressure slightly higher than the pusher pressure was ap-
plied to the RPV with Valve D open to force all of the
filtrate lodged between the valve stem and the RPV top
into the vessel. The RPV was brought back slowly to
room temperature, vented, and weighed.
The strontium level in the diluted filtrate was then
detennined by atomic absorption using a Varian Model
775ABQ. The amount of soluble SrS04 in the test brine
at the test temperature and pressure was calculated.
Solubility Data
Data from the first phase of this study, the Box-Behnken
design shown in Fig. 1, are presented in Table 3.
"Trial" refers to the point on the design cube in Fig. 1
and "Run" refers to the actual experiment perfonned.
Each trial was duplicated or triplicated. Since it was im-
possible to achieve the exact target test pressure and
temperature, only the actual experimental values are
295
 TABLE 4-SrSO 4 SOLUBILITY DATA-FACTORIAL DESIGN

SrS0 4
Experimental Predicted
p Temperature NaCI Salinity SrS0 4 (mg/L)
Trial Run (psig) (OF) (gIL) (J;) (mg/L) Eq.1 Eq. 2
14 65237 806 155 13.9 0.489 322 274 315
65250 811 140 13.9 0.489 324 294 333
65254 811 140 13.9 0.489 337 294 333
15 65236 2,290 155 13.9 0.489 358 288 323
65253 2,268 140 13.9 0.489 366 312 340
65249 2,281 140 13.9 0.489 363 312 340

16 65228 854 238 13.9 0.489 227 186 230

65245 825 238 13.9 0.489 217 185 230

17 65229 2,310 238 13.9 0.489 277 197 240

65243 2,336 238 13.9 0.489 258 197 240

18 65252 818 140 114 1.399 750 775 760

65255 818 137 114 1.399 787 780 763
65258 825 137 114 1.399 787 780 763

19 65239 2,351 155 114 1.399 778 800 762

65251 2,309 140 114 1.399 796 823 777
65247 2,329 140 114 1.399 633 824 778

20 65231 942 238 114 1.399 615 656 654

65246 859 238 114 1.399 596 654 654

21 65244 2,390 238 114 1.399 608 696 680

65256 2,377 235 114 1.399 701 699 683

listed in Table 3. The square root of the total ionic
strength listed includes the contribution from the soluble
SrS04' The solubility is reported in the table as mg/L
SrS04'
Regression analyses were performed on the data
shown in Table 3. Many methods of defining the
variables and the solubility were tried to determine the
best fit. Use of the expressions log Ksp, pressure (p) in
psia, inverse temperature (T) in K, and the square root of
the total ionic strength, Jii, provided the best regression
results. The regression equation that best fits the data on
Table 3 is
logk sp =A+BJ!;. +Cp.+Dp
=EIT+F(J!;.IT), .................... (1)
where
A = -10.379,
B = 4.245,
C = -0.715,
D = 3.5x 10- 5 ,
E = 1271.894, and
F = -656.441.
Regression Eq. 1 was used to predict a solubility for each
run in Table 3 and is listed in the next-to-last column.
Table 4 shows the results of the solubility determina-
tions made on the 2 3 factorial design shown in Fig. 1
along with solubilities predicted from the regression on
the Box-Behnken design (Eq. 1). A new regression study
296
was run on the combination of the Box-Behnken and fac-
torial designs and yielded an equation of the same form
as Eq. 1 with
A = -10.325,
B = 4.274,
C = -0.732,
D = 4.6x 10- 5 ,
E = 1257.736, and
F = -667.119.
Very little change in the coefficients was observed.
The solubility data from the previous work 4 (Table 5)
that falI within the design of this study [i.e., the 104 and
160°F, 0 to 200 giL (40 and 71°C, 0 to 200 g/dm 3 )
NaCl] were combined with the data of this study. The
212°F (l00°C) was omitted because flashing of water
appeared to have occurred during the open filtration by
the method used then. Regression analyses were per-
formed on the combined data. It was found that a model
similar to Eq. 1 would not provide a satisfactory fit.
Several methods of curve fitting were attempted. Dif-
ferent variable definitions and combinations were
studied. Several forms of expressing the dependent
variable also were examined. The best equation yielded
the smalIest difference between experimental and
predicted SrS04 solubilities.
The final mathematical model or regression equation
derived for strontium sulfate solubility was
X
logK sp = - ............................... (2)
R
SOCIETY OF PETROLEUM ENGINEERS JOURNAL
 1000
900
800
700
600

TABLE 5-SrSO 4 SOLUBILITY OAT A (0 psig)· 500

SrS0 4
-701=
Experimental Predicted • 400 ··············1041=
(mg/L) 0
Temperature NaCI Salinity SrS0 4 !(!
--1601=
til
(OF) gIL (J;) (mg/L) Eq.1 Eq.2 ..J
104 0.00 0.0177 143 133 153 a
~
300
104 1.00 0.1325 186 174 202
104 5.02 0.2946 272 246 280
104 10.00 0.4152 342 309 344
104 20.00 0.5867 470 410 439
104 50.75 0.9339 671 627 626
104 100.00 1.3106 830 794 784 200
104 150.00 1.6050 896 811 851
104 200.00 1.8530 868 739 861

160 0.00 0.0152 106 87 104

160 1.00 0.1325 151 118 147
160 5.02 0.2946 222 172 218
160 10.00 0.4152 290 222 277
160 20.00 0.5867 384 305 368
160 50.75 0.9339 572 504 555 100
160 100.00 1.3106 731 692 717 1 2 3
160 150.00 1.6050 776 754 787
160 200.00 1.8530 798 726 798
\I Total Ionic Strength
• Adapted from Ref. 4. Fig . 5-SrSO 4 solubility vs. ionic strength.

where
x = liT,
R = A+BX+C.J;, +DJ.I.+Ep2 +FXp+G.J;,p, and
p = pressure [psig (kPa)].
The coefficients of the equation were
A = 0.266498 x 10 -3,
B = -244.828xW- 3 ,
C = -0.191065 X 10 3 ,
D = 53.543xW- 6 ,
E = -1.383xW- 12 ,
F = I.W3323xW- 6 , and
G = -0.509x 10- 9 .
The last two columns in Tables 3, 4, and 5 contain the
predicted SrSO 4 solubility using both the initial model
(Eq. 1) and the final model (Eq. 2). Correlation coeffi-
cients were calculated between the experimental values
and the calculated values for both Eqs. 1 and 2. The
coefficients were 0.9857 and 0.9918, respectively, in-
dicating the improvement in fit achieved by the use of
Eq.2.
Fig. 5 shows plots of Ksp vs. Jil. for three tempera-
tures. The solid lines are Ksp values predicted from Eq.
2. The X's are values from work by Fletcher et al. 8 The
first plot at 75 OF (24°C) represents extrapolated values
APRIL 1983
from the data base reported in this paper. The last two
plots fall within the temperature range of this work, but
are extrapolated beyond the ionic strength of 3.5. A
significant difference between Fletcher's work and this
work is the brine composition. We used pure NaCI brine
compared with Fletcher's use of brines containing
various amounts of calcium, sulfate, magnesium, and
other ions. The agreement between the plots and data
from the work of Fletcher appears good, deviating for
those values extrapolated from this work.
Discussion
The SrS04 solubility data, as presented in the tables, in-
dicate that pressure changes from 100 to 3,000 psig (689
to 20 684 kPa) result in only very small differences in
solubility. The trend of solubility decreasing with
temperature is obvious and is similar to that observed in
the earlier work 4 graphically presented in Fig. 6. In-
creasing ionic strength results in a significant increase in
SrSO 4 solubility.
A better understanding of solubility trends becomes
more apparent after regression studies are performed.
The regression performed on just the Box-Behnken data
(Trials 1 through 13) resulted in Eq. 1, which provides a
reasonable model for the surface of the cube design. It
does not provide as precise a model in the interior of the
cube design. This becomes apparent by comparing the
predicted solubilities from Eq. 1 with the experimental
297
 25

20

15

10

25

-
0
~
UI
20

-><
Q. 15

1000.0 _------r-- ~
1/1

10

800.0

::::r 800.0
a
~
25

d 400.0
re
III
20

200.0
15
0.0

10

Fig. 6-SrSO 4 solubility in NaCI brines. 4 Fig. 7-SrSO 4 solubility vs. temperature and ionic strength
at 1,550 psig (Eq. 2).

ones in Tables 4 and 5. Use of a model with the same
algebraic fonn as Eq. 2, with just Box-Behnken design
data, did provide a slight improvement.
When all the experimental data were used in the
regression study and the algebraic fonn of Eq. 2 was
employed, a significant improvement in the prediction of
solubility within the Box-Behnken interior was ob-
served. The predicted solubilities presented in the last
columns of Tables 3, 4, and 5 are from the final model
using Eq. 2.
A three-dimensional, computer-generated presentation
of the solubility of SrS04 as a function of temperature
and square root of ionic strength at 1,550 psig (l0 687
kPa) is presented in Fig. 7. The surface indicates a
decrease in solubility with increasing temperature and a
significant increase with increasing ionic strength. The
solubility reaches a maximum in highly concentrated
brines, similar to that seen in Fig. 6. This behavior is
observed in all temperatures and pressures within the
design of this study. Surfaces similar to the one in Fig. 7
can be generated at any pressure within the design. The
298
contours remain the same across the entire range of the
design with only minor shifts along the solubility edge of
the cube. Any point on the surface in Fig. 7 represents a
saturated solution. Points above it represent super-
saturated solutions, while those below it represent sub-
saturated solutions. The model presented can predict
potential scaling problems. This is done by comparing
the predicted solubility product (Ksp) to the actual prod-
uct of the strontium concentration times the sulfate con-
centration (Q), and is written
mg/L Sr mg/L SO 4 6
Q= X xlO- ............ (3)
87.62 96.0
If the ratio QI K sp is greater than 1. 0, then a potential
scaling problem exists.
The infonnation needed to do these computations
typically comes from water analyses. If we know the ion
concentration, we can substitute the appropriate values
into Eq. 3 to solve for Q and into Eq. 4 to solve for ionic
SOCIETY OF PETROLEUM ENGINEERS JOURNAL
 ......
. ~.:.:.:::.:.:::.:::.......
1000

900 /'

"...........
/ ... .....,
800
/;......
/.:..,
",
700
.,,?'
.. 800
."l
Il
0
en
a::
en
500
/'
t/
TABLE 6-SrSO 4 SCALE FORMATION PREDICTION'

Chemical
..J
......
CJ
400 1/
/f
---CICI,

:::E /i .............. 50:1 mg/l CI:g/l NaCI

,
(mg/L) A B C Ii
Sr 560 565 550 300
Ii - - - - NaCI
Ii
Na 43,817 39,292 44,275 Ii _. _. - 100:1 mgll Ca:gll NaCI
Ca 7,930 6,977 6,967 if
Mg 1,422 1,270 1,215
CI 85,555 76,148 83,957 Ii
S04 210 540 100 If
200
HC0 3 232 311 360 If
pH 7.0 7.3 7.5 Ii
Temperature, OF 230 230 230
Pressure, psig 2,500 2,500 2,500
~ 1.6441 1.553 1.617
Log(Ksp) -4.8623 -4.8165 - 4.8200
Ksp 1.37x10- 5 1.52x 10- 5 1.51x10-5
0 1.39 x 10- 5 3.63x10-5 0.65 x 10- 5
OIK sp 1.01 2.38 0.43 100L-----~-------7------_L______~----
Scale
prediction borderline scale prone no scale
Field
observation borderline scale no scale
Fig. 8-SrSO 4 solubility vs. ionic strength in different brines
• Adapled from Eq. 2. (250C). 4

strength appears to be the limiting solubility at a given ionic

1L=lh~mizi2, ........................... (4)
where mi is the molality of the ith ion and Zi is the
charge of the ith ion. Having calculated ionic strength,
and knowing temperature (11K) and pressure [psig
(kPa)], Eq. 2 can be used to solve for log (K sp )' If the
ratio QI K sp is equal to or less than 1.0, then no scaling is
predicted. If that ratio is greater than 1.0, then the water
is supersaturated and scaling is possible. The greater the
ratio of QI K sp' the more likely scaling will occur.
Examples of the calculation used to make scale predic-
tions are shown in Table 6. The analyses of three waters
from the same field are listed in the table. From the
water analyses, ionic strengths were calculated. The
calculated ionic strength, temperature [230°F or
(0.OO261/K)], and pressure [2,500 psig (17237 kPa)]
were substituted into Eq. 2 to generate a predicted log
(K sp )' Q was calculated on the measured strontium and
sulfate ion concentration, and the values of Q were com-
pared to the respective predicted Ksp values. At the bot-
tom of the table are listed the field observations. Brine A
came from a well that had borderline SrSO 4 scaling
tendencies. Brine B was from a well that had SrS04
scale when untreated with scale inhibitor, and Brine C
was from a well that had no scaling problem. These
observations corresponded well with calculated ratios
(Q1Ksp) which were 1.01, 2.38, and 0.43, respectively.
As stated in the earlier work,4 the pure NaCI system
APRIL 1983
strength. In complex brines containing calcium or
magnesium, the apparent solubility of SrS04 is greater
than NaCI brine for the equivalent ionic strength. 4,8 One
explanation of the difference is the formation of ion
pairs, for example, SrS04 0, MgS04 0, CaS04 0. Most
analytical methods (atomic absorption and titrimetric
methods, for example) do not distinguish between free
ions and ion pairs. A second possibility is the precipita-
tion of CaSO 4 resulting from the presence of a large con-
centration of calcium ions. Such a precipitation would
keep the sulfate ion concentration low and permit a
greater amount of SrS04 to dissolve. Complex ions have
been dealt with in numerous publications such as
Davies 10 and Nancollas. II The magnitude of the devia-
tion from expected solubility is indicated by the work of
Jacques et al., 4 from data generated by Davis and Col-
lins,9 and from the work of Fletcher et al. 8 Fig. 8 is a
plot of data from Jacques et al. showing mg/L SrS04 as
a function of ionic strength. The brines contain different
levels of calcium. The pure calcium brine shows the
largest deviation from the pure NaCI system of 20% at
an ionic strength of about 0.7 at 7r F (25 °C). 4 Davis
and Collins measured solubilities 10 to 20% higher in
calcium and magnesium brines than in the pure NaCl
brines of the same ionic strength. In most oilfield brines,
NaCI will be the largest source of ions, while
magnesium and calcium are present in lesser quantities
(for example, see water analyses in Table 6). The
bivalent metal concentration would affect the apparent
299
solubility of SrSO 4, but the expected magnitude would
be small. Fletcher et al. 8 showed this in their work with
complex brines (see Fig. 7).
It should be understood that the results presented in
this paper deal with thermodynamic equilibrium. While
solubility products are useful in the prediction of scaling
tendencies, they do not account for kinetic factors, such
as the presence of seeding crystals or surfaces, rates of
water production, physical changes (such as temperature
and pressure) during production, and other factors that
influence the rate of precipitation.
Conclusions
This study has resulted in the development of an easy-to-
use method for determining solubilities of salts under
downhole conditions. It has also resulted in a model
useful for predicting SrS04 scaling potential in oilfield
systems.
Nomenclature
K sp = molar solubility product of SrSO 4
p = pressure, psia (kPa)
Uthmaniyah Field," paper SPE 9626 presented at the 1980 SPE
Middle East Oil Technical Conference, Manama, Bahrain, March
9-12.
3. Essel, A. and Carlberg, B.L.: "Strontium Sulfate Scale Control
by Inhibitor Squeeze Treatment in the Fateh Field," 1. Pet. Tech.
(June 1982) 1302-06.
4. Jacques, D.F., Gollberg, J.B., and Whiteside, W.e.: "Strontium
Sulfate Solubility and the Effects of Scale Inhibitors," paper
presented at the NACE Middle East Corrosion Conferencel79,
Bahrain, April 15-17.
5. North, N.A.: "Pressure Dependence of SrS04 Solubility,"
Geochim. Cosmochim. Acta (1974) 28,1075.
6. MacDonald, R.W. and North, N.A.: "The Effects of Pressure on
the Solubility of CaCO 3' CaF 2' and SrSO 4 in Water," Cdn. 1.
Chem. (1974) 52, 3181.
7. "Strategy of Experimentation," E.!. DuPont de Nemours and Co.
short course, Wilmington, DE (Oct. 1975).
8. Fletcher, G.E., French, T.R., and Collins, A.G.: "A Method For
Calculating Strontium Sulfate Solubility," U.S. DOEI
BETC/RI-801l0 (1981).
9. Davis, J.W. and Collins, A.G.: "Solubility of Barium and Stron-
tium Sulfates in Strong Electrolyte Solutions," Env. Sci. and
Tech. (1971) 10,1039.
10. Davies, e.W.: Ion Associations, Butterworths Pub. Inc.,
Reading, MA (1962).
II. Nancollas, G.H.: Interactions In Electrolyte Solutions, Elsevier
Sci. Pub. Co. Inc., New York City (1966).

T = temperature, K SI Metric Conversion Factors

Z = pressure, psig (kPa) OF (OF-32)/1.8
p., = total ionic strength of (OF+459.67)/1.8
L x 1.0* E+OO
References micron x 1.0* E+OO
I. Nassivera, M. and Essel, A.: "Fateh Field Sea Water Injec- psig x 6.894 757 E+OO
tion-Water Treatment, Corrosion, and Scale Control," paper ·Conversion factor is exact. SPEJ
SPE 7765 presented at the 1979 SPE Middle East Oil Technical
Original manuscript received in Society of Petroleum Engineers office Dec. 21, 1980.
Conference, Manama, Bahrain, March 25-29. Paper accepted for publication Oct. 4, 1982. Revised manuscript received Dec. 1,
2. Stoffer, K.G. and Lindlof, J.C.: "Strontium Sulfate Wellbore 1982. Paper (SPE 9625) first presented at the 1981 SPE Middle East Oil Technical
Scaling-A Case Study of Seawater Injection Compatibility in the Conference held in Manama, Bahrain, March 9-12.

300 SOCIETY OF PETROLEUM ENGINEERS JOURNAL