Modeling of Tubing Pickling in Conjunction
With Acidizing Treatments
Saleh H. Al-Mutairi, SPE, Saudi Aramco, and A. Daniel Hill, SPE, and
Hisham A. Nasr-El-Din, SPE, Texas A&M University
Summary
Pickling well tubulars to prevent pumping unwanted materials into the
formation is an issue that should be decided on an individual-job basis.
Acid pickling of tubing is a process of flow in a tube or an annulus
with heterogeneous reactions occurring at the wall of the tube. The
reacting solid species at the wall are primarily mill scale and other
types of inorganic scale. The fluids injected for pickling are usually
strong acid solutions [hydrochloric acid (HCl)] but also may include
surfactants, organic solvents, and gelled solutions to aid in lifting sol-
ids from the wellbore. Thus, the process is a complex one involving
several reactions and perhaps multiple stages of fluids.
Tubing pickling is an essential part of well-stimulation treat-
ments if the main acid is to be bullheaded. Standard design param-
eters for a pickle treatment depend on experience and personal
judgment. It appears that the standard pickle-treatment design is
overestimating the required volume of pickling acid. Field data
indicated that excessive acid volumes were used for tubing pickling
because large returns of unreacted (live) acid were recovered at
the surface. Careful analysis of flowback samples obtained from
several deep gas wells showed that only a fraction of the acid was
consumed by mill scale and other tubing contaminants.
In this paper, proposed mechanisms are presented to explain
the behavior of acid contact with the tubing and a mathematical
model is developed to predict acid consumption and dissolution of
tubular contaminants. The model applies to the bullheading case
where the acid is pumped down the tubing and then flowed back
to the surface. The model considers the reaction of acid with mill
scale (Fe3O4). The equations formulated are solved numerically to
predict the concentrations of major chemical species in the well
flowback samples.
Such a model is extremely valuable in optimizing future pick-
ling treatments. The acid volume needed for pickling operations
can be reduced significantly, and other improvements can be made
without extensive and costly field testing. Finally, recommenda-
tions are given to design acid-pickling treatments better.
Introduction
The main purpose of the chemical stimulation treatment is to remove
formation damage and enhance well productivity. The treatment nor-
mally consists of acids, solvents, surfactants, gels, or combinations
of these chemicals. When these treating fluids are pumped through
the injection or production tubing, they may squeeze contaminants
from the tubing into the formation, which may plug the perforations
or damage the formation (McLeod 1984). Therefore, the purpose of
the tubing-pickling treatment is to clean the injection or production
tubing of damaging material before conducting the main stimula-
tion treatment.
In general, a tubing-pickling treatment is a prerequisite if treat-
ing fluids, such as the acid in acidizing treatments, or the fluid
transporting sand particles, as in gravel packing, are to contact
both the tubing and the formation during the main treatment. The
main purpose of tubing pickling is to remove mill scale, rust, iron
sulfide, and pipe dope from the tubing.
Copyright © 2010 Society of Petroleum Engineers
This paper (SPE 93291) was accepted for presentation at the SPE International Symposium
on Oilfield Chemistry, The Woodlands, Texas, USA, 2–4 February 2005, and revised for
publication. Original manuscript received for review 15 May 2008. Revised manuscript
received for review 18 November 2009. Paper peer approved 21 October 2009.
822
During acidizing, acid must reach the formation at its original
strength and contain as few contaminants as possible for the job
to have maximum effectiveness. Pickling the tubing before acid
jobs removes scale and other contaminants and allows successful
treatment. In hydraulic fracturing, the scouring action of the prop-
pants on the tubing mixes mill scale, pipe dope, and other tubular
contaminants into the main fracturing fluids during proppant
transportation. These materials can be introduced into the forma-
tion and may reduce the conductivity of the fracture. In addition,
these materials are capable of damaging the formation or plugging
the perforations and, therefore, must be removed from the tubing
before pumping the main treatment fluids.
Description of the Process
In a tubing-pickling treatment, several tubular contaminants are
reacting with the acid as it flows through the pipe. Reactions of
these contaminants with the acid are classified as heterogeneous
reactions because they are liquid/solid reaction systems. Fig. 1
depicts the way the acid enters, reacts with, and exits the pipe. Part
A shows the system before the acid enters the pipe. Once the acid
enters the pipe, as indicated in Part B, it starts reacting with the
inner surface of mill scale, forming a slug of spent acid (reaction
products) in front of the live acid. As the acid travels farther down
the pipe, as in Part C, it is consumed by the mill scale, creating a
larger spent-acid slug. Part D shows the displacement fluid push-
ing the acid forward in the pipe. The volume of the live-acid slug
decreases, while that of the spent-acid slug increases as the acid
travels farther down the tubing. That is because the front of live
acid is converted to spent acid by the reaction.
A careful review of actual field data supports the idea of this
mechanism. Fig. 2 shows the total iron concentration (main reac-
tion product) in the acid front, mainly in the region occupied by the
spent acid (Nasr-El-Din et al. 2002b). Because the chloride ion is
not consumed in the reaction, it can be used as a tracer. Therefore,
we can identify the region between the live-acid and the chloride-
ion curves as the spent-acid slug (Fig. 3).
Figs. 2 and 3 suggest that only the front of the live acid is
reacting with mill scale. Hence, if the whole slug of live acid is
reacting, then high levels of iron would appear over the whole
interval. It is important to note that the amount of mill scale in a
well is limited. Only a very thin layer of scale (0.01 in. thick) is
available for the reaction (Smith 1990). As described, mill scale is
removed during the reaction with the early flow of acid. The fresh
acid that comes later is in excess and does not find scale to react
with until the acid advances to unreacted scale down the tubing.
The fresh acid is inhibited with corrosion inhibitors, which prevent,
or at least reduce, the dissolution of well tubulars.
The preceding has described the macroscopic picture of the
problem by considering the whole system. In Fig. 4, a microscopic
visualization demonstrates the proposed mechanism of the hetero-
geneous reaction between the acid and the reactants on the inner
surface of mill scale. Two main processes control the reaction (Wil-
liams et al. 1979; Loewen et al. 1990; Nasr-El-Din et al. 2002b):
• Mass transfer—This process includes the transfer of the react-
ing species from the bulk of the acid to the surface of the mill scale
and then the transfer of the reaction products from the surface of
the scale back into the bulk stream.
• Reaction kinetics—The rate of the reaction between the acid
and mill scale that coats the inner surface of the tubing.
September 2010 SPE Journal
 A. The original fluid inside the pipe. No acid has come yet. Flowback
Direction

120000 24

B. The original fluid is being pushed ahead of the acid. Spent

100000 Fe (Total) 20

HCl Concentration, wt%

Fe Concentration, mg/L
acid appears between the live acid and the original fluid. HCl
80000 16

60000 12

C. More acid is reacted. Larger spent acid slug is formed. 40000 8

20000 4

0 0
0 50 100 150
D. The original fluid exited the pipe. Displacing fluid pushing
Flowback Time, min
the acid.
Fig. 2—Fe and HCl concentrations in well flowback samples
for Well A.

where [H+], k, and  are the concentration of hydrogen ion, the

Original Live Acid reaction-rate constant, and reaction order, respectively.
Fluid Eq. 2 indicates complete dependency of the reaction rate on the
concentration of the hydrogen ion. However, this dependency is not
Spent Acid Displacement
absolute (Blesa et al. 1994). Other investigators showed that the
Fluid
reaction rate is affected slightly by the chloride concentration and
the pH of the solution (Galus and Adams 1963). The reaction rate is
first order at low acid concentrations and of a higher order at high
Fig. 1—Description of the flow stages inside tubing. acid concentrations (Blesa et al. 1994). Different values for the
reaction-rate constant were reported in the literature (Table 1).

Removal of Mill Scale Scale Thickness

Magnetite (Fe3O4) is one of many iron oxides forming rust. It forms
on the steel surfaces as a result of corrosion reactions (exposure to
oxygen and water). Magnetite has high solubility in mineral acids.
It reacts with HCl according to Eq. 1 (Smith 1990):
8 HCl + FeO·Fe2O3 = FeCl2 + 2 FeCl3 + 4 H2O. . . . . . . . . (1)
The rate of reaction, rHCl, of the acid with magnetite can be repre-
sented by Eq. 2 (Blesa et al. 1994):
−rHCl = k[H+], . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (2)
The chemical and physical characteristics of the iron oxide and iron
sulfide scales were found to depend on temperature, pressure, and
pH. For iron sulfide scale, the scale density and thickness depend
on the depth and age of the scale (Nasr-El-Din and Al-Humaidan
2001). Magnetite scale forms a thin layer on the surface of the
tubular as a result of cooling the tubular during manufacturing
or as a result of corrosion reactions. The thickness of the scale is
estimated to be 0.01 in. A 0.003-in. hard and dense layer covers
the surface of new pipes, with a 0.007-in. softer and flaky layer
on top of the hard layer (Smith 1990). The foregoing description
was given for new well tubulars. Thus, it can be assumed that the

1.2 125000
HCl
1 Chloride
100000
Dimensionless HCl and Cl –

Fe
Iron Concentration, mg/L

0.8
75000
0.6
50000
0.4

25000
0.2

0 0
0 20 40 60 80 100 120 140
Flowback Time, min

Fig. 3—HCl, Cl−, and Fe dimensionless concentrations for Well A.

September 2010 SPE Journal 823

 Inside Surface of the scale
Acid Scale Thickness

MILLSCALE LAYER
Outer Radius of the Tubing
Transport to the scale surface

Transport from tubing surface ao : Inner Radius of Scale

Reacon at scale surface

Fig. 4—Steps of acid reaction with mill scale (Williams et al.

1979).
Fig. 5—A cross section of scaled tubing.

scale is uniform throughout the tubing. However, Smith (1990) tives on the reaction rate were not considered when developing the
claimed that the soft layer (0.007 in.) may pop off the pipe during model. However, the presence of the corrosion inhibitor was used
straightening and handling. Once a new pipe is placed in a well, to support the idea of minimal attack on the base metal by the acid.
higher temperatures will enhance the corrosion rate and the scale Water-wetting surfactants removed any residual hydrocarbons and
film becomes thicker. The nonuniform temperature along the enhanced acid/mill-scale contact.
tubulars makes the growth rate of the scale layer nonuniform as In order to visualize the problem, consider a cross section of the
well. Therefore, it is expected that the thickness of mill scale will pipe as shown in Fig. 5. Before the acid contacts the scale, it has an
increase with depth. original radius a0. The radius then increases to a new larger value
a as the acid stays in contact with the scale for a time increment
Development of a Mathematical Model t. The volume of magnetite dissolved during this time interval is
When the acid enters the pipe, it reacts with a thin layer of magne- given by (Al-Mutairi 2004)
tite deposited on the inner surface of the scale. Therefore, the inside 
diameter of the scale layer becomes relatively larger as the acid ∂a kCHCl  MWHCl
= . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (3)
dissolves more magnetite. The amount of dissolved magnetite and, ∂t Fe3O4
hence, the effect on the inner diameter of the scale layer increases
as the acid contact time with the pipe increases. Thus, there exists Using appropriate dimensionless transformations, Eq. 3 is con-
a critical contact time at which all the magnetite will be consumed verted to
and extra exposure time will not be beneficial.
Several assumptions were made in developing the mathematical ∂ 
= DK HCl , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (4)
model. It was first assumed that the effect of the acid on the base ∂
metal of the pipe was minimal. This was because of the presence of
a sufficient amount of proper corrosion inhibitor in the live acid. On where , HCl, and  are dimensionless groups for the radius, acid
the basis of this assumption, radius enlargement caused by subse- concentration, and time, respectively.
quent live-acid contact with the base metal was neglected. Further,
isothermal conditions were assumed during acid pumping and well kL
D= . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (5)
flowback. In other words, the variation in temperature with depth a0u
was neglected. Also, temperature variations caused by the heat of
the reaction were neglected. Acid concentration was a function of and
the axial position and time only. There was no radial change in the
(C )

acid concentration. Therefore, the problem was described by axi- 0
HCl  MWHCl
symmetric cylindrical flow. This assumption was well justified K= . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (6)
Fe3O4
because of the turbulent-flow conditions that were present during
the acid-pickling treatments. The mill scale was assumed to be
distributed evenly over the surface of the tubing. In other words, the Eq. 4 describes the change of the radius at each axial position
scale thickness was independent of the axial position in the pipe. in the pipe as the acid reacts with the scale. One known condition
Transport of reactants to the inner surface of the scale or reaction is that the radius of the scale, at a certain point, will not change
products by diffusion was accounted for by numerical dispersion. until the acid reaches this point. Thus,
Mixing that might be induced by produced gases during flow-
back was not considered in the model. This mixing was noted for ε >  ,  ( , ε ) = 1 , . . . . . . . . . . . . . . . . . . . . . . . . . . (7)
in the flowback analysis of a few wells, and its effect on model
predictions was not significant. Finally, the effects of acid addi- where ε represents a dimensionless length.

TABLE 1—REACTION-RATE CONSTANT FOR MAGNETITE WITH HCl

Oxide Form Solvent Temp. (°C) K (m/s) Reference

–6
Carbon paste 0.1 M HCl 25 (7.4–10) 10 Tanaka and Tamamushi (1964)
–6
Powder 2 M HCl 70 6.95 10 Cumper et al. (1982)
–5 –4
Carbon paste HCl+C1 – 8.5 10 –1 10 Galus and Adams (1963)

824 September 2010 SPE Journal

 The dimensionless group, D, given in Eq. 5 is the Damköhler ⎧1, during pumping the acid
number, which represents the ratio of the rate of acid consumption  Fe ( , 0 ) = ⎨ . . . . . . . . . . . . . . (17)
to the rate of acid convection. Eq. 6 represents the acid capac- ⎩0, otherwise
ity number, which is defined as the ratio of the amount of scale
dissolved by the acid occupying a unit volume of the pipe to the The chloride ion will not react inside the tubing. Therefore, the
amount of scale present in that volume. mole-balance equation for the chloride is similar to that of HCl
A differential mole balance on the reacting HCl describing the with the reaction term omitted. A differential mole balance on the
flow and reaction inside the tubing can be derived to obtain chloride ion inside the tubing can be derived to give

 ∂CCl − ∂CCl −
∂CHCl ∂C 2 kCHCl = −u . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (18)
= − u HCl − . . . . . . . . . . . . . . . . . . . . . . . . . . (8) ∂t ∂x
∂t ∂x a
Using appropriate dimensionless transformations, Eq. 18 is
Using appropriate dimensionless transformations, Eq. 8 is trans- transformed to
formed to
∂ Cl − ∂ Cl −
∂ HCl ∂ HCl   + = 0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (19)
+ + DM HCl = 0 , . . . . . . . . . . . . . . . . . . . . . . . (9) ∂ ∂ε
∂ ∂ε 
The following initial and boundary conditions accompany Eq.
where 19:

 Cl ( 0, ε ) = 0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (20)
(
⎡ C0  ⎤ )
M = ⎢ 2 0 ⎥ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (10)
HCl

⎢ CHCl ⎥
⎣ ⎦
⎧1, during pumping the acid
 Cl ( , 0 ) = ⎨ . . . . . . . . . . . . . . . (21)
The first term in Eq. 9 represents the accumulation of the acid at ⎩ 0, otherwise
any point in the tubing. The second term describes the transport
of the acid through the tubing. And the third term represents the Eq. 19 is a linear, first-order partial-differential equation. It
depletion of the acid by the reaction with mill scale. describes the propagation of chloride ion in the wellbore during
Because there is no acid present initially in the tubing, we pickling treatment. Eqs. 9 and 14 are nonlinear first-order partial-
have differential equations. This system of equations was solved using
numerical techniques. Space derivatives were approximated using
 HCl ( 0, ε ) = 0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (11) central finite differencing, whereas backward finite differencing was
used to approximate time derivatives. This approximation method is
called upwind differencing. The accuracy of this method depends on
Also, the concentration of the acid at the tubing inlet is constant the values of  and ε. It is known that the solution of the equations
during pumping the acid. Thus, using this method is stable as long as (/ε) < 1. It should be noted
that the values of  and ε used in the present study were 0.0002
and 0.0005, respectively. This means that /ε was 0.4.
⎧1, during pumping the acid
 HCl ( , 0 ) = ⎨ . . . . . . . . . . . . . . (12)
⎩ 0, otherwise Field Application
Tubing-pickling treatments of two wells (A and B) were used to
In a similar way, the mole balance on the total iron gives validate this model. The two wells were completed with 5½-in. 20#
NK-3SB C-95, low-carbon steel tubing. The tubing was 10,850 and

10,040 ft long for Wells A and B, respectively, and both wells had a
∂CFe ∂C ⎛ 3 ⎞ kC 4.778-in. inner diameter. Other pertinent data are given in Table 2.
+ u Fe − 2 ⎜ ⎟ HCl = 0 . . . . . . . . . . . . . . . . . . . . . (13)
∂t ∂x ⎝ 8⎠ a Acid additives included a corrosion inhibitor and a water-wetting
agent. More details on the acid formulation were given by Nasr-El-
Using appropriate dimensionless transformations, Eq. 13 is trans- Din et al. (2002b). It is important to note that the gases produced
formed to from these wells contained hydrogen sulfide. Therefore, a hydrogen
sulfide scavenger was added to the acid. More details on hydrogen
 sulfide scavengers were given in Nasr-El-Din et al. (2002a).
∂ Fe ∂ Fe  In both wells, pickling fluids were pumped down through the
+ − DG HCl = 0 , . . . . . . . . . . . . . . . . . . . . . . . . (14)
∂ ∂ε  production tubing until the leading edge of the acid was 50 ft
above the upper set of perforations. Then, the acid was flowed back
where immediately through the same path to the surface using reservoir
pressure. A total of 4,000 gal of 20 wt% HCl was used to pickle the
tubing of Well A (a perforated gas well). The estimated flowback
( )

⎡6 C0 ⎤
G=⎢
HCl
⎥ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (15) rate for Well A was 1.7 bbl/min, while the acid pumping rate was
⎢ 8 CFe0 ⎥ 3 bbl/min. Flowback rate was based on the time required to flow
⎣ ⎦ back the 4,000 gal of acid. Well B was pickled with only 2,000 gal
of 20 wt% HCl. It was also a perforated gas well. The estimated
As with Eq. 9, Eq. 14 has the accumulation and transport terms.
flowback rate for Well B was 3.7 bbl/min, while the acid pumping
The last term in Eq. 14 represents the loading of iron, which bal-
rate was 3 bbl/min. Flowback rate was based on the time required
ances the depletion-of-the-acid part in Eq. 9. Eq. 15 represents
to flow back the 2,000 gal of acid.
the stoichiometry between the acid and the iron in the chemical
After pickling Well A, samples of well flowback were collected
reaction (Eq. 1). The set of initial and boundary conditions for the
once per minute for the first 37 minutes. Then the well was shut in
total iron can be given as
for 17 minutes because of operational problems, and no samples were
collected. Sampling was resumed once the well was put on production,
 Fe ( 0, ε ) = 0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (16) and sampling was conducted every 3 to 4 minutes. This relatively short

September 2010 SPE Journal 825

 TABLE 2—FIELD DATA FOR THE TWO EXAMPLES
Variable Well A Well B Units
3
Acid volume 15.14 7.57 m
3
Acid conc. 6,016 6,016 mol/m
Acid M.W. 36.5 36.5 g/mol
3
Iron conc.* 0.895 0.896 mol/m
6 6 3
Scale density 5 10 5 10 g/m
–4 –5
Scale thickness 2.26 10 3.33 10 M
Scale M.W. 231.4 231.4 g/mole
Dissolving power 0.7925 0.7925 g Fe3O4 /g HCl
–3 –3 3
Pumping rate 4.59 10 9.71 10 m /s
Fluid velocity 0.397 0.839 m/s
Injection time 3,299 780 seconds
Tubing radius 0.0607 0.0607 m
Tubing length 3,292 3,060 m
* Iron concentration in the pumped acid.

Results and Discussion

TABLE 3—DIMENSIONLESS PARAMETERS
FOR THE TWO WELLS
Variable Well A
D 1.901
K 0.035
G 2,520
sampling interval was needed to follow acid reactions with wellbore
contaminants. Well B was pickled and flowed back with no operational
problems. Sampling was taken once per minute for 13 minutes.
The well flowback samples typically contained condensate, the
spent acid, and solids. Each phase was separated and analyzed.
Several analyses were conducted on the aqueous phase, including
measuring sulfide, pH, Cl, Na, Ca, Mg, Zn, Pb, Mn, and total iron.
Sulfide content and pH were measured on site. The concentrations
of various cations were determined by inductively coupled plasma
emission spectroscopy. The chloride ion was measured by titration
using a 0.1N silver nitrate solution. The concentration of HCl was
measured by acid/base titration using a 0.1N NaOH solution.
More than 15 vertical gas wells were examined in this study. Model
predictions were in good agreement with field data. However, we
Well B will discuss model predictions and field data for Wells A and B,
1.923 which represent the trends noted with other wells.
0.035
Well A. The thickness of the scale for Well A was estimated from the
2,520 treatment data by integrating the total-iron-concentration curve. The total
amount of iron removed from the tubing was found to be 1027 kg. This
amount of iron was divided evenly on the total inner surface area of the
tubing to obtain a scale thickness of 0.009 in. The amount of manganese
was calculated in a similar manner to obtain 6.6 kg. On the basis of this
information, the amount of metal dissolved by acid per unit area was
0.0098 lbm/ft2, which was far below the maximum allowable value of
0.05 lbm/ft2 (Nasr-El-Din and Metcalf 2008). This validates our assump-
tion of having an effective corrosion inhibitor. Details of these calcula-
tions were given in Nasr-El-Din et al. (2002b).
Table 3 lists values of dimensionless variables for the pickling
treatment that was conducted on Well A. The values of D, K, and
G were calculated to be 1.901, 0.035, and 2,520, respectively. The
value of the reaction order, , was taken as unity. Figs. 6 through 11

1.2
Squares: Actual Chloride Concentration
Solid Line: Modeled Chloride Concentration
Dimensionless Cl– Concentration, ψCl

0.8

0.6

0.4

0.2

0
0 0.5 1 1.5 2
Dimensionless Flowback Time, θ
Fig. 6—A comparison of actual and predicted dimensionless chloride concentration for Well A.

826 September 2010 SPE Journal

 1.2 2500

Dimensionless Iron Concentration, ψFe

Chloride Iron

Dimensionless Cl– Concentration, ψCl

1
2000

0.8
1500
0.6
1000
0.4

500
0.2

0 0
0 0.5 1 1.5 2
Dimensionless Flowback Time, θ
Fig. 7—Predicted dimensionless concentrations of iron and chloride for Well A.

show comparisons of model predictions with field data. It should
be noted that, when bullheading the acid, the fluid that was pumped
last would flow back first. This means that the acid front appeared
toward the tail end of the acid slug.
The dimensionless chloride concentration is shown in Fig. 6 as
a function of dimensionless time. The dimensionless chloride con-
centration was almost zero in the initial stages of well flowback.
Then, it reached unity at a dimensionless time of 1.2. Finally, the
chloride concentration gradually decreased and exhibited some-
what of a tailing effect in the very last stage of well flowback.
A good agreement between the model predictions and field data
was obtained at dimensionless time of 1.2. However, model predic-
tions were not as close to the field data for the very last stage of
flowback. This difference is mainly because of the flow of gases
during well flowback. The flow of gases through wellbore fluids
during well flowback increased mixing and dispersion of the acid
slug. This resulted in more dispersion and tailing of the profile
of the chloride ion. The high mixing and dispersion effects were
escalated by the 17 minutes of shut-in time during well flowback.
However, the effect of gases produced during well flowback on the
maximum iron and amount of iron dissolved by the acid was not
significant, as will be shown in the following.
The way the fluids are pumped in and out of the wellbore
affected the time when various species were produced. Fig. 7
shows the dimensionless concentrations of chloride (tracer) and
iron ions as a function of the dimensionless time. The iron peak
was expected to appear in the flowback samples toward the end of
the acid slug. The chloride ion was noted first in the well flowback
samples and was followed by iron.
Fig. 8 shows the dimensionless acid concentration as predicted
by the model and obtained from field data. Reaction occurs at
late times where the acid contacted the tubing first. Fig. 9 shows

1.2
Circles: Actual HCl Concentration
Dimensionless HCl Concentration, ψHCl

Solid Line: Modeled HCl Concentration

1

0.8

0.6

0.4

0.2

0
0 0.5 1 1.5 2
Dimensionless Flowback Time, θ
Fig. 8—Predicted and actual dimensionless acid concentration for Well A.

September 2010 SPE Journal 827

 2500
Squares: Actual Iron Concentration
Solid Line: Modeled Iron Concentration

Dimensionless Iron Concentration, ψFe

2000

1500

1000

500

0
0 0.5 1 1.5 2
Dimensionless Flowback Time, θ
Fig. 9—Predicted and actual dimensionless iron concentration for Well A.

the actual and predicted dimensionless iron concentrations. The
actual and simulated maximum iron concentrations were 105 488
and 107 203 mg/L, respectively. The actual and simulated total
amounts of iron produced in the well flowback were 1027 and
1040 kg, respectively. The chloride ion was used to track the
emergence of acid from the tubing because field data were sampled
randomly. Then, proper alignment was performed to match the
field and model data. Simply, the actual time was converted to
dimensionless time.
The results shown in Figs. 6, 8, and 9 indicate that the model
predicted field data very well. The effect of mixing caused by the
presence of gases produced during well flowback was found to be
not significant in the wells examined.
Finally, the model was used to optimize the acid-pickling treat-
ments. Two main parameters were examined: the reaction-rate
constant and the acid pumping rate. Parameters of Well A (Table
2) were used in the optimization process.
Fig. 10 and Table 4 show the maximum iron concentration and
total amount of iron removed as a function of the reaction-rate con-
stant. Reaction-rate constants greater than 1×10−5 m/s will ensure
removing all of the mill scale from the tubing. On the other hand,
lower reaction-rate constants do not provide good cleaning with the
current amount of acid. Matching the field data with model predic-
tions indicates that the best value for the reaction rate constant is
6.95×10−6 m/s, which is in good agreement with the data given by
Cumper et al. (1982).

160000 1200

1000
Maximum Iron Concentration, mg/L

120000
800
Amount of Iron, kg

80000 600

400
40000
200
Maximum Iron
Concentration
0 0
1×10–7 1×10–6 1×10–5 1×10–4 1×10–3 1×10–2
Reaction-Rate Constant , m/s
Fig. 10—Effect of reaction-rate constant on the maximum iron concentration for Well A.

828 September 2010 SPE Journal

 150000

120000

Maximum Iron Concentration, mg/L

90000

60000

30000

0
0 2 4 6 8 10 12

Flowrate, bbl/min

Fig. 11—Effect of the acid pumping rate on maximum iron concentration for Well A.

The effect of the acid pumping rate on the iron peak is examined
in Fig. 11 and Table 5. The data show that the iron peak decreases
when the acid pumping rate is increased. This result is reasonable
because slow rates would give more contact time and allow the
acid front to be loaded with more iron. It is clear that the iron peak
increases with increasing the dimensionless parameter D (Eq. 5).
This parameter increases with increasing the reaction-rate constant
or by decreasing the acid pumping rate.
Fig. 12 shows the effect of the acid pumping rate on the amount
of iron dissolved by the acid (the cleaning process). High pumping
rates do not give ample contact time for all the scale to be removed.
However, for this example, flow rates as high as 4 bbl/min still
give good results.
In field application, the practical limitations for the acid pump-
ing rate (maximum = 5 bbl/min) can ensure sufficient contact time
for scale removal. This is indicated by the flat section of the iron
curve shown in Fig. 12. This section means that the contact time
provided by 4 bbl/min was sufficient to remove all the scale and
that more contact time provided by the lower flow rate was not
needed. If the volume of acid were increased, then the flat section
would cover even higher flow rates because the contact time would
increase. For example, with 8,000 gal of acid, the flat section in
Fig. 12 would cover up to 7 bbl/min.
Well B. The thickness of the scale for Well B was estimated from
the treatment data by integrating the total-iron-concentration curve.
The total amount of iron removed from the tubing was found to
be 140.5 kg. This amount of iron was divided evenly on the total
inner surface area of the tubing to obtain a scale thickness of 0.001
in. The difference in the amount of the dissolved iron is because
the tubing of Well A is longer and older.
Table 3 lists values of dimensionless variables for the pickling
treatment that was conducted on Well B. The values of D, K, and
G were calculated to be 1.923, 0.035, and 2,520, respectively.
The shorter length of the tubing of Well B and the increase of the
pumping rate caused the parameter D for Well B to be different
from that of Well A. The values of K and G did not change because
they reflected the properties of the fluids used and scale, which
did not change. Figs. 13 through 15 show comparisons of model
predictions with field data for Well B.
The dimensionless chloride concentration is shown in Fig. 13
as a function of dimensionless time. A good agreement between
the model predictions and field data was obtained at dimensionless
TABLE 5—EFFECT OF ACID PUMPING RATE
ON IRON REMOVAL, WELL A

Flow Rate Maximum Iron Amount of Iron

TABLE 4—EFFECT OF REACTION-RATE CONSTANT (bbl/min) Concentration (mg/L) Removed (kg)
ON SCALE REMOVAL, WELL A
0.5 125,589 1,035
Reaction Rate Maximum Iron Amount of Iron 1 121,040 1,040
Constant (m/s) Concentration (mg/L) Removed (kg)
1.7 107,203 1,040
–7
1 10 3,451 51.48 2.5 92,270 1,039
–6
1 10 30,206.4 456.34 3.5 76,851 1,014
–5
1 10 117,716.7 1,040.4 4.5 65,421 943
–4
1 10 126,085.6 1,020.6 5.5 56,801 853
–3
1 10 126,080.4 962.2 7 47,237 715
–2
1 10 126,079.5 960.5 10 35,400 535

September 2010 SPE Journal 829

 1200

1000

Iron Removed, , kg
800

600

400

200

0
0 2 4 6 8 10 12

Flowrate, bbl/min

Fig. 12—Effect of the acid pumping rate on the amount of iron removed for Well A.

time of 1.05. Well B was not shut in during the flowback, as was acid from the tubing because field data were sampled randomly.
the case with Well A; therefore, the mixing and dispersion effects Then, proper alignment was performed to match the field and
of the acid slug were not noticed. As a result, a tailing profile of the model data. Simply, the actual time was converted to dimension-
chloride ion did not appear. The shock-like profile at both ends of less time.
the curve indicate that the viscous pills that were used to separate The results shown in Figs. 13 through 15 indicate that the model
the acid slug from other stages were effective. predicted field data very well. The way the fluids are pumped in
Fig. 14 shows the dimensionless acid concentration as predicted and out of the wellbore affected the time when various species were
by the model and obtained from field data. Reaction occurs at late produced. Fig. 16 shows the dimensionless concentrations of the
times where the acid contacted the tubing first. Fig. 15 shows the HCl, the chloride, and the iron as a function of the dimensionless
actual and predicted dimensionless iron concentrations. The actual time. As was discussed earlier and shown in Figs. 2 and 3, the
and simulated maximum iron concentrations were 90 436 and iron peak appeared in the flowback samples toward the end of the
90 699 mg/L, respectively. The actual and simulated total amounts acid slug. The loss of the acid was substituted by a gain in iron
of iron produced in the well flowback were 145.5 and 140.6 kg, concentration. The model gave good explanation of the pickling
respectively. The chloride ion was used to track the emergence of process, which agreed with our interpretation in Fig. 2.

1.2

Squares: Actual Chloride Concentration

Dimensionless Chloride Concentration, ψCl–

1
Solid Line: Modeled Chloride Concentration

0.8

0.6

0.4

0.2

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6

Dimensionless Flowback Time, θ

Fig. 13—A comparison of actual and predicted dimensionless chloride concentration for Well B.

830 September 2010 SPE Journal

 1.2
Circles: Actual HCl Concentration

HCl
1 Solid Line: Modeled HCl Concentration

Dimensionless HCl Concentration, ψ

0.8

0.6

0.4

0.2

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Dimensionless Flowback Time, θ

Fig. 14—Predicted and actual dimensionless acid concentration for Well B.

Comparison between the model and the actual results measured time for reaction to dissolve the scale from tubing totally. However,
in the field is summarized in Table 6. It is shown that the model some extra acid is recommended as an allowance. The allowance
predicted the final outcome of the pickling treatments accurately. should provide enough acid to exit with the same acid strength.
The difference between model prediction and actual field data The following formula is recommended to calculate the volume
varied from one case to another, with a maximum of 10%. It is of 20 wt% HCl acid (V in gallons) required to remove magnetite
believed that the main source of this difference is the simplifying from the tubing or the liner:
assumptions made in the development of the model.
V = 4.6874 × ts × L × F(d + ts), . . . . . . . . . . . . . . . . . . . . . (22)
Procedure To Estimate Acid Volume
On the basis of the results obtained in the present study, the fol- where d = diameter of tubing or liner in inches, F = safety fac-
lowing procedure is recommended to determine the acid volume tor, L = length of tubing or liner in feet, and ts = thickness of the
required to remove the scale. scale in inches.
The required volume of acid should be calculated from the There is a maximum acid pumping rate that should not be
reaction stoichiometry between scale and the acid. This applies exceeded. This rate should give enough contact time for efficient
to current average pumping flow rates, which give ample contact cleaning of the tubing. This rate depends on the volume of the acid

2000
Dimensionless Iron Concentration, ψFe

Squares: Actual Iron Concentration

1600
Solid Line: Modeled Iron Concentration

1200

800

400

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6

Dimensionless Flowback Time, θ

Fig. 15—Predicted and actual dimensionless iron concentration for Well B.

September 2010 SPE Journal 831

 1.2 2000

Modeled Chloride Concentration

1 Modeled HCl Concentration

Dimensionless Chloride and HCl Concentration

1600
Modeled Iron Concentration

Dimensionless Iron Concentration, ψFe

0.8

1200

0.6

800

0.4

400
0.2

0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6

Dimensionless Flowback Time, θ

Fig. 16—Predicted dimensionless concentrations of chloride, HCl, and iron for Well B.

TABLE 6—PARITY TABLE FOR MODELED CASES

Cases Model Actual % Difference

Amount of iron for Well A 1,040 1,027 1.27

Amount of iron for Well B 140.6 145.5 3.37
Amount of iron for Well C 56.6 62.4 10.24
Maximum concentration of iron for Well A 107,203 105,488 1.63
Maximum concentration of iron for Well B 90,699 90,436 0.29
Maximum concentration of iron for Well C 29,799 26,995 9.54

used. Exceeding the maximum rate will give insufficient contact • The pickling process was found to be affected by acid pump-
time and some of the scale will not be removed. For Well A, as ing flow rate and the reaction-rate constant (i.e., by Damköhler
noted in Fig. 12, flow rates as high as 4 bbl/min are permissible number).
without affecting the results. • A slight deviation between model predictions and field results
Finally, the best way to use this model is first to calculate the was noted in Well A because of mixing produced gases with spent
theoretical amount of acid needed to remove all the scale using acid. However, the effect of this on the amount of iron produced
Eq. 22. Second, identify the maximum flow rate that is planned and on the maximum iron concentration noted in the flowback
for the treatment. Then, use this flow rate with an amount of acid samples was not significant.
that is greater than the one calculated until the dimensionless acid
concentration in the flowback samples reaches unity. The amount Nomenclature
of scale removed from the tubing is independent of flow rates less a = inner radius of scale, m
than the one entered into the model. Therefore, unplanned lower a0 = initial inner radius of scale, m
flow rates will not affect the pickling process. CCl − = chloride-ion concentration, mol/m3
CCl0 − = initial chloride-ion concentration, mol/m3
Conclusions
CFe = total iron concentration, mol/m3
In this work, a mathematical model was developed to predict the CFe0 = initial total iron concentration, mol/m3
acid and total iron concentrations following acid-pickling treat-
CHCl = HCl concentration, mol/m3
ments when the acid is bullheaded. The following conclusions can
be drawn from this work:
0
CHCl = initial HCl concentration, mol/m3
• Acid-pickling treatment removed significant amounts of iron. There- d = diameter of tubing or liner, in.
fore, it is strongly recommended to pickle well tubulars before acid D = Damköhler number
treatments, especially when the acids will be bullheaded. k = reaction-rate constant, m/s
• Model predictions were in a good agreement with field results. K = acid capacity number

832 September 2010 SPE Journal

 L = length of tubing, ft
MWHCl = molecular weight of HCl, g/mol
rHCl = reaction rate of HCl, mol/m2·s
t = time, seconds
ts = thickness of scale, in.
u = fluid velocity, m/s
V = acid volume, gal
x = axial position along the tubing, m
 = reaction order
 = dissolution power, g scale/g HCl
ε = dimensionless space increment
 = dimensionless time increment
ε = dimensionless tubing length, x/L
 = dimensionless time, tu/L
 = dimensionless pipe radius, a/a0
 Cl − = dimensionless chloride concentration, CCl − CCl0 −
Fe = dimensionless iron concentration, CFe/C 0Fe
HCl = dimensionless HCl concentration, CHCl/C 0HCl
Acknowledgments
The authors wish to acknowledge Saudi Aramco for granting permission
to publish this paper. Udhailiyah Lab Unit, SALD, and the Chemistry
Unit and Advanced Instruments Unit of the Research and Development
Center, Saudi Aramco, were very helpful in their field monitoring and
analysis of the spent acids, respectively. Members of the stimula-
tion group, Research and Development Center, are acknowledged for
processing well flowback samples. All engineers in Gas Production
Engineering Division and local service companies are acknowledged
for many useful discussions. The authors would like to thank the anony-
mous reviewers of this paper for many useful comments.
References
Al-Mutairi, S.H. 2004. Modeling of Tubing Pickling in Conjunction With
Matrix Acidizing Treatments. MSc thesis, University of Texas at Aus-
tin, Austin, Texas.
Blesa, M.A., Morando, P.J., and Regazzoni, A.E. 1994. Chemical Dissolu-
tion of Metal Oxides, 285. Boca Raton, Florida: CRC Press.
Cumper, C.W.N., Grzeskowiak, R., and Newton, P. 1982. Effect of pyrrole,
indole and their dimethyl derivatives on the dissolution of magnetite. Cor-
rosion Science 22 (6): 551–557. doi: 10.1016/0010-938X(82)90054-3.
Galus, Z. and Adams, R.N. 1963. The Investigation of the Kinetics of
Moderately Rabid Electrode Reactions Using Rotating Disk Electrodes.
J. Phys. Chem. 67 (4): 866–871. doi: 10.1021/j100798a036.
Loewen, K., Chan, K.S., Fraser, M., and Leuty, B. 1990. A Well Stimulation
Acid Tube Clean Methodology. Paper CIM 1990-047 presented at the
Petroleum Society of CIM Annual Technical Meeting, Calgary, 10–13
June. doi: 10.2118/1990-047.
McLeod, H.O. Jr. 1984. Matrix Acidizing. J Pet Technol 36 (12): 2055–
2069. SPE-13752-PA. doi: 10.2118/13752-PA.
Nasr-El-Din, H.A. and Al-Humaidan, A.Y. 2001. Iron Sulphide Scale:
Formation, Removal and Prevention. Paper SPE 68315 presented at the
International Symposium on Oilfield Scale, Aberdeen, 30–31 January.
doi: 10.2118/68315-MS.
Nasr-El-Din, H.A. and Metcalf, A.S. 2008. Workovers in Sour Environ-
ments: How Do We Avoid Coiled Tubing (CT) Failures. SPE Prod &
Oper 23 (2):112–118. SPE-87622-PA. doi: 10.2118/87622-PA.
Nasr-El-Din, H.A., Al-Humaidan, A.Y., Fadhel, B.A., Frenier, W.W., and
Hill, D.G. 2002a. Investigation of Sulfide Scavengers in Well-Acidiz-
ing Fluids. SPE Prod & Fac 17 (4): 229–235. SPE-80289-PA. doi:
10.2118/80289-PA.
Nasr-El-Din, H.A., Al-Mutairi, S.H., and Al-Driweesh, S.M. 2002b. Les-
sons Learned From Acid Pickle Treatments of Deep/Sour Gas Wells.
Paper SPE 73706 presented at the International Symposium and Exhibi-
tion on Formation Damage Control, Lafayette, Louisiana, USA, 20–21
February. doi: 10.2118/73706-MS.
Smith, B. 1990. Proper Treatment of Tubulars Key to Iron Control. Pre-
sented at the 37th Annual Southwestern Petroleum Short Course, Lub-
bock, Texas, USA, 18–19 April.
September 2010 SPE Journal
Tanaka, N. and Tamamushi, R. 1964. Kinetic parameters of electrode
reactions. Electrochimica Acta 9 (7): 963–989. doi: 10.1016/0013-
4686(64)85045-3.
Williams, B.B., Gidley, J.L., and Schechter, R.S. 1979. Acidizing Funda-
mentals. Richardson, Texas: SPE.
Dr. Saleh H. Al-Mutairi is a researcher with the stimulation group
in the Operations Services Division at Saudi Aramco EXPEC
Advanced Research Center. e-mail: saleh.mutairi.1@aramco.
com. His research interests include well stimulation, formation
damage, tubular corrosion problems, and rheology of fluids.
He holds a BS degree in chemical engineering and an MBA
degree, both from King Fahd University of Petroleum and
Minerals, Dhahran, Saudi Arabia. He holds an MS degree in
petroleum engineering from The University of Texas at Austin.
He holds a PhD degree in petroleum engineering from Texas
A&M University. Al-Mutairi has 10 years of experience in well
stimulation operations and research. He has several published
work in this area and is an active member of SPE. Dr. A.D. Hill is
a professor and holder of the Robert L. Whiting Endowed Chair
in Petroleum Engineering and associate head of the Petroleum
Engineering Department at Texas A&M University. Previously,
he taught for 22 years at The University of Texas at Austin after
spending 5 years in the industry. He holds a BS degree from Texas
A&M University and MS and PhD degrees from The University of
Texas at Austin, all in chemical engineering. He is the author
of the SPE monograph Production Logging: Theoretical and
Interpretive Elements, coauthor of the textbook Petroleum
Production Systems, coauthor of the SPE book Multilateral Wells,
and author of more than 130 technical papers and five pat-
ents. He has been an SPE distinguished lecturer, has served on
numerous SPE committees, and was founding chairman of the
Austin SPE section. He was named a distinguished member of
SPE in 1999 and received the SPE Production and Operations
Award in 2008. He currently serves on the SPE Editorial Review
Committee and is Chairman for the SPE Hydraulic Fracturing
Technology Conference. Hill is an expert in the areas of produc-
tion engineering, well completions, well stimulation, production
logging, and complex well performance (horizontal and mul-
tilateral wells), and has presented lectures and courses and
consulted on these topics throughout the world. Dr. Hisham A.
Nasr-El-Din is a professor and holder of the Stephen A. Holditch
Faculty Fellowship at Texas A&M University in petroleum engi-
neering. e-mail: hisham.nasreldin@pe.tamu.edu. Previously, he
worked for 15 years as Principal Professional and Team Leader
of the Stimulation Research and Technology Team, EXPEC ARC,
Saudi Aramco. Before joining Saudi Aramco, he worked for four
years as a staff research engineer with the Petroleum Recovery
Institute in Calgary. He also worked as a research associate
with the University of Saskatchewan, the University of Ottawa,
and the University of Alberta, all in Canada. His research inter-
ests include well stimulation, formation damage, cementing,
drilling fluids, two-phase flow, enhanced oil recovery, rheol-
ogy, conformance control, interfacial properties, adsorption,
and nondamaging fluid technologies. Nasr-El-Din has several
patents and has published more than 400 technical papers.
He is an adjunct professor with the University of Alberta and
has supervised several MS and PhD students. He has received
numerous awards within Saudi Aramco for significant contri-
butions in stimulation and treatment-fluid technologies and
stimulation design, and for his work in training and mentoring.
Nasr-El-Din holds BS and MS degrees from Cairo University and
a PhD degree from the University of Saskatchewan, Canada,
all in chemical engineering. He serves on the SPE steering
committees on corrosion and oilfield chemistry, is a review
chairperson for SPE Journal and is a technical editor for SPE
Production and Operations and SPE Drilling and Completion
journals. He was invited to give keynote presentations in vari-
ous SPE and NACE conferences. He received the SPE Regional
Technical Discipline Award for Production and Operations in
2006, was named a Distinguished SPE Member in 2007, and
received SPE awards for Outstanding Associate Editor (SPE J.)
and Outstanding Technical Editor (SPE Prod & Oper) in 2008. In
addition, he received SPE awards for Outstanding Associate
Editor (SPE J.) and the SPE Production and Operations in 2009.
833