SPE

SPE 15176
Carb( n Dioxide Corrosion in Oilwell Cements
by R,A, Bruckdorfer,Dowell Schlumberger
SPE Member

Copyrlghl 1986, Socielyof Petroleum Engineers

This paper was prepared for presentation al the Rocky Mountain Reg!ona, Meeting of the Socte!y of Petroleum Engineers held m B!llmgs, MT, May
19-21, 1986.
This paper was selected forpresentalion byan SPE Program Commlfleefollowlng rewewof inlormatlon contained m anabstract stibm!tledbylhe
author(a). Ccmlenls of the paper, as presented, have not been rewewed bythe Socie!yof Petroleum Engineers and are aubjecl tocorrection by the
author(s). The ,naterial, as presented, does not necessarily reflect any position of the Society of Pelroleum Engineers, 11sofficers, or members. Papera
presanledaf SPE meefings aresubject topublicatlon review by Editorial CommMeesof the Socletyof Petroleum Engineers. Permmontocopyia
restricted to an abstract of not more than 300 words. Iiluslrat[ons may not be copied, The abstract should contain conspicuous acknowledgment of where
and bv whom the oaper is presented. Write Publicafiona Manager, SPE, p.o. BOX833836. Richardson, TX 75083-3836. Telex, 730989. SpEDAL.

ABSTRACT
Carbon dioxide corrosion studies of oil-well
portland cements were initiated using a new microsample technique to determine the effect of carbonic
acid on portland cement slurry formulations and to
develop a high carbonic acid corrosion-resistant
cement for carbon dioxide Enhanced Oil Recovery applications.
Earlier results from studies using two-inch API
cement cubes showed carbonic acid had @essentially no
effect on cement after relative short test periods at
elev.?ed temperatures.
Similar results with two-inch
API cement cubes also were reported in recent literature.
Because carbonic acid corrosion in cements was
difficult to observe and measure using two-inch API
cement cubes, a new microsample technique was developed. Use of this new technique, which represents an
accelerated testing method, showed oil-well cements
undergo a rapid deterioration in a wet carbon dioxide
environment. Similar tests with two-inch cubes showed
essentially no cement deterioration under the same
conditions.
Experimental details of this new microsample
technique
are discussed.
Data relating cement
strength loss and carbon dioxide penetration depths to
cement type and
slurry
formulation
are
reviewed.
Included in the discussion is a new cementing formulation which shows significant promise as a high carbondioxide-resistant, oil-well cemwt.

INTRODUCTION

have

Carbon dioxide Enhanced Oil Recovery applications

seen a surge of activity in the last several

years.
Of the 40 projects that are estimated to be
under way, 20 are in Texas and 7 are in the Gulf Coast
region, while the remaining are distributed throughout
the midwest and western states and Canada.
Although the carbon dioxide Enhanced Oil Recovery
process and carbon dioxide corrosion in oil and gas
production are well documented in the literature, l
very little published information is available on
:arbonic aciG corrosion in oil-well cements.2
How?ver, the corrosion of cement structures by the leaching action of the carbon-dioxide-laden waters is well
documented in the literature.3
It is well known that
:arbon-dioxide-laden water can reduce hydrated portIand cement to a soft, amorphous, silica gel.q
The basic chemistry describing this process is as
follows.
(1)

C02+H20#

(2)

CaOHz+H+HCOs+CaCOs

(3)

C-S-H + H + HC03 +CaCOs

H2C03#H++HC03-

+ amorphous silica

(I),
approximately 1% of the dissolved
In Step
carbon dioxide reacts with water to form carbonic
acids
As the carbon-dioxide-laden water diffuses
into the cement matrix, the dissociated acid is free
to react with the calcium hydroxide, which makes Up
20% of the cement composition, and the hydrated calcium silicates (Steps (2) and (3), respectively). If
the reaction would stop after the initial carbonation
of the calcium hydroxide, the cementitious calcium
carbonate formed would cause an increase in compressive strength.
Although a strength increase is
observed, as more carbon-dioxide-laden water invades
the matrix, several new equilibria are established.

References and illustrations at end of paper.

1

(4)

COz+HzO

+CaCOs

(5)

Ca(HCOs) z+

X~Ca(HCOs)z

Ca(OH)z#2CaC03+

H20

IN OIL-WELL CEMENTS

CARBON DIOXIDE CORROSIl

In the presence of excess carbon dioxide (Step

4), the calcium carbonate is converted to watersoluble calcium bicarbonate.
As more carbon-dioxideladen water enters the matrix, the equilibrium is
moved to the right.
The end result is more watersoluble calcium bicarbonate which can diffuse out of
the matrix. As the calcium bicarbonate forms, another
equilibrium reaction is established (Step 5).
The
dissolved calcium bicarbonate reacts with more calcium
hydroxide
to
form calcium carbonate and fresh
water. The importance of Step (5) is that the fresh
water formed in the cement matrix is now available to
dissolve more calcium bicarbonate. Providing there is
a continuing supply of carbon-dioxide-laden water, the
reactions will continue to leach cementitious material
from the matrix.
As the process continues, an inThe end
crease in permeability and porosity occurs.
result is an amorphous silca gel that lack; structural
Downhole, this equates to a loss of
integrity.5
casing corrosion protection and zonal isolation.
Because of the potential problems that are inherent in cements in a wet carbon dioxide environment,
such as that found around a well in a carbon dioxide
Enhanced Oil Recovery WAG project, this study (
initiated to investigate the effects of wet carbon
dioxide on oil-well cements and to determine if portland cements could be made more resistant to carbonic
acid attack.

EXPERIMENTAL
Sample Preparation
To study the effects carbonic acid has on hydraulic portland cements, three sample geometries were
In
Investigated over the course of the experiments.
the first phase, two-inch API cubes were used, whereas
the second phase utilized one-inch-diameter cylindriBased on the data gained in the first
cal samples.
two phases, sample geometry was changed to microcylindrical . The end result was that carbonation effects
were observable over relative short exposure periods
compared to cubic and macrocylindrical cement samples.
Because the use of cubic cement samples was
limited to one data set, only the preparation of the
macro- and microcylindrical samples is discussed.
The l-in. OD x 2-in. [2.54-cm OD x 5.05-cm]
macrosamples were cast using a 50-cc disposable plastic syringe as a mold.
Before preparing the cement
slurry and casting the sample, the plunger was removed
and the syringe tip was sealed.
All cement slurries
tested were prepared according to API Specification
10, Section 5, using distilled water.
Before mixing
the slurry, a defoaming agent was added to the mix
water to reduce air entrainment.
The cement samples
were cast by slowly pouring the slurry down onc side
of the syringe barrel, then puddling the cement with a
stirring rod (API Specification 10, Section 5) to
remove any trapped air.
After sealing the syringe
with a rubber stopper, the samples were cured for 72
hr at 3,000 psi [20.68 MPa] and 175F [79.4C].
The
samples were demolded by cutting the syringes tip
e~d, placing the mold in warm water to expand the
plastic and pushing the cured cement cylinder out of
the barrel with a syringe plunger.
Microcylindrical cement samples were prepared
from 2-in. API cubes which were machined and cored to
produce a 0.275-in. OD x 0.5-in. [7.O-mm OD x 12.7-mm]

SPE 15176

cylinder.
To make a set of samples, the cement systems were prepared as described above and cast into 2in. [5.08-cm] API cubes. Curing conditions were 72 hr
at 3,000 psi [20.68 MPa] and 175F [79.4C].
P.fter
demolding the samples, each cube was machined into a
2-in. x 2-in. x l/2-in. [5.D8-cm x 5.08-cm x 1.27-cm]
cement blank by first removing about a l/2-in.
[7.0mm] thick section from the top and bottom of each cube
to form a rough 2-in. x 2-in. x l/2-in. [5.08-cm x
5.08-cm x 1.27-cm] blank, then milling the major faces
flat and parallel to each other,
After sealing the
faces with a suitable polymer to eliminate possible
end effects, nine 0.275-in. 00 x 0.5-in. [7.O-mm OD x
12.7-mm] samples were cored from each blank.

SANPLE CARBONATION
Cement carbonation was done under static conditions using a modified water-filled autoclave.
Although both static and dynamic conditions exist downhole, most of the cement sheath at the formation
interface is experiencing an essential static carbon
dioxide exposure. The only exception is the immediate
area around the perforations which is under a dynamic
staie during both gas and water injection.
The main
advantage to dynamic testing is that carbonic acid
movement to the surface of the cement sample is not
dependent on diffusion through water.
Because the
rate of carbonation in cement is diffusion controlled,
moving the carbon-dioxide-laden water over the sample
most probably accelerates the corrosion rate by replenishing the carbon dioxide at the cement surface,
but is not necessarily representative of actual downhole conditions.
Hence, the simpler approach, i.e.,
static carbonation, was selected for the cement carbonation experiments.
Cement samples were carbonated by loading them in
the autoclave in a sample holder, After filling the
chamber with water and sealing, liquid carbon dioxide
was injected into the water using a sparge tube connected to a carbon dioxide tank until 900 psi [6.21
MPa] was recorded.
The test chamber was recharged
ever two to three days over the four- to six-week
carbonation period.
Temperature and pressure were
maintained at 175F [79.4C] and 3,000 psi [20.68
MPa].
To eliminate any effects temperature and pressure had on the results, a second autoclave was used
for a control sample set. This was run under the same
conditions as the test set except that the chamber was
not charged with carbon dioxide.

CARBONATION MEASUREMENTS
Carbonic acid corrosion effects on the macrosample sets were determined using compressive strength
and carbonation penetration depth measurements, whereas only compressive strength measurements were used
for the microsample sets.
Compressive strengths of
the l-in. OD x 2-in. [2.54-cm OD x 5.08-cm] samples
were measured by first cutting off 0.5 in. [1.27 cm]
of each cylinder end, then milling the end surfaces
perpendicular to the cylinders axis.
This was done
to remove any deteriorated cement at the sample ends
and to ensure that the cylinder was perpendicular tc
the instruments ?latens, both of which cause scattering in the test results.
Compressive strengths
were determined using a Baldwin instrument.
Because
the length-to-diameter ratio was less than two, the
resultant strengths were multiplied by 0.91 to obtaifl

532

R. A. BR[ (DORFER

SPE 15176
corrected
68).

compressive strengths (ASTM Standard C42 -

The depth the carbonic acid penetrated into

cement was determined by staining the freshly cut
of a cylinder with a 1% phenolphthalein solution.
measuring the unstrained area, which indicates
calcium hydroxide is present in the cement matrix,
depth of carbonic acid invasion into the matrix
determined.

the
end
By
no
the
was

Microsample compressive strengths were determined

on an Instron 1123 instrument.
Prior to the use of
research showed that providing the
this method,
length-to-diameter ratio of the cement microcylinder
was greater than a given value, the compressive
strength measured was equal to that obtained from a 2in. [5.08-cm] cement cube according to API Specification 10, Section 7, when using the Instron compresAt least three samples
sive/extension test machine.
were tested, and the average strength was reported for
Since
each cement formulation per sampling period.
each cylinders ends were protected with a polymer
of pseudo-infinite
coating,
creating
a cylinder
length, effects which cause data scattering were
essentially eliminated.

RESULTS AND DISCUSSION

Results from the first two phases of this study
indicated that the cement mass of a macrosample, i .,
2-in. [5.08-cmJ cubes and l-in. diameter [2.54-cm
diameter] cylinders, was too large for the relative
short carbon dioxide exposure period to show significant cement deterioration in the samples. Compressive
strengths of the carbonated cements compared to the
control samples showed no decrease in strength or
observable
deterioration on
the cement
surface.
Carbonic acid penetration depth into the cement matrix, although measurable, was in all cases insignificant.
As an example, Figures 1 and 2 show compressive
strength of a Class H cement and a 50% fly ash/Class H
cement as a function of slurry density. Comparison of
the carbonated curves to the control curves shows
essentially, within the experimental error of the
method, no decrease in strength after a 31-day carbonation period at 175F and 3,000 psi [79.4C and
Although no change in compressive
20.68 MPaJ.
strength was observed, carbonation depth measurements
showed that as cement density increased carbon dioxide
penetration decreased (Table 1). It is interesting to
note that in all cases, except for the 17-lb/gal
[2,037-kg/m31 systems, the carbon dioxide penetration
depth for the fly ash/Class H systems was three times
greater than the straight Class H cements.
Similar
results were obtained when Class A and C cements were
compared to a Class H cement mixed at 15.6 lb/gal
[1,864 kg/m3] (Table 2).
There was no significant
difference in carbon dioxide penetration rates between
the three systems.
Comparison of the compressive
strength of the carbonated samples to those of the
control, in all cases, showed no decrease after an
exposure period of 31 days at 175F and 3,000 psi
[79.4C and 60,68 MPa] .
Although carbonation was
observed in the samples, it was concluded that to get
meaningful results the carbon dioxide exposure period
would have to be increased or the cement sample mass
The latter of the two choices was chosen
reduced.
because sam~le turnaround would be relatively fast

compared to that projected for the macrosamples.

Based on the carbon dioxide penetration rates in the
macrosamples, a 0,275-in. x 0.5-in. [7.0 mm-x 12.7-mm]
cylindrical samples geometry was calculated to undergo
complete carbon dioxide penetration wit,hin a four- to
six-week exposure period at 175F and 3,000 psi
[79.4C and 20.68 MPa].
To determine if the conclusion ,~,xiafrom the
macrosample data was correct, a set of cement microsamples was prepared and tested (Set No. 1). Results
from the initial test set showed that oil-well cements
deteriorate in the presence of carbonic acid and the
use of microsamples represented a pseudo-accelerated
method of testing the corrosion resistance of cement
formulations to carbonic acid attack. Test conditions
were 175F and 3,000 psi [79.4C and 20.68 MPa].
Exposure period was six weeks with sampling periods at
three and six weeks.
All cement slurries were mixed
at 15.O-lb/gal [1,797-kg/m3]
density.
The results
were reported as percent strength loss compared to the
compressive strength of the control sample set.
As
shown in Table 3, hydraulic portland cements deteriorate in the presence of carbonic acid.
8oth Class H
and Class C cements underwent an 80% strength loss
after six weeks of exposure time. As was indicated in
the macrosample carbon dioxide penetration data,
increasing the
cement density,
i.e.,
increasing
cement-to-water ratio, decreased the rate of carbon
dioxide corrosion (Table 4).
Changing the cement to
water from 1.88 to 2.36 by weight improved the corro25% (Table
sion resistance of the cement by dbOUt
Because high cement-to-water ratio slurries
4).
require a dispersant to improve pumping, the effect a
dispersant would have on the corrosion rate was
studied. As indicated by the data (Table 5), addition
of up to 1% dispersing agent by weight of cement has
essentially no effect on the carbon dioxide corrosion
rate.
Salt containing cement slurries were also
investigated in the study. The initial data indicate
that addition of up to 37% salt has no effect on the
cements resistance to carbon dioxide attack compared
to that of a neat cement slurry of the same density
(Table 6).
In all cases, about an 80% strength loss
resulted from carbonic acid attack.
In view of the
findings of this study which indicated that carbon
dioxide corrosion of the casing cement sheath downhole
may be real and pose zonal isolation and casing corrosion problems over the life of the well, a second
experiment was initiated using the microsample method.
The objective of the second set of experiments
was to determine how various additives affect carbonic
acid resistance in cements and to develop a high
carbon dioxide resistant oil-well cement for carbon
dioxide Enhanced Oil Recovery.
Marynell Class H
cement, which is common to the Permian Basin area, was
used in all cement systems.
Except for the density
study, all cement slurries were mixed at 16.O-lb/gal
Carbonation was at 3,000 psi
[1,917-kg/m3] density.
and 175F [20.68 MPa and 79.4C] for an exposure
period of six weeks. Results were reported as percent
strength loss as compared to the control samples which
were run under the same condition, except in the
absence of carbon dioxide.
~~CF!USe
Of several Shutdowns during carbonation, each set of experiments,
i.e., Set
No. 1 and Set No. 2, should be considered
independent of each other and only relative comparisons made between the two data sets to avoid introducing any determinate errors into the study.
Sampling periods were at zero, three and six weeks.
As
was shown in the last study, increasing the cement-to-

&,

CARBON
-. . . DIOXIDE
- . -. .- -- CORROSION IN OIL-WELL CEMENTS

water ratio increased the carbon dioxide resistance of

the cement (Table 7). As cement density was increased
from 15.0 to 17 lb/gal [1,917 to 2,037 kg/m3], percent
strength loss went from 100% to 54%. Although there
is a small apparent decreasing difference in carbonic
acid resistance between the 16- and 17-lb/gal [1,917and 2,037-kg/m3] slurries, within the experimental
error of the method, increasing the cement density
from 16,0 to 17.0 lb/gal [1,917 to 2,037 kg/m31 M no
real effect on the carbonic acid resistance of the
The
cement based on a six-week exposure period.
A 3%
opposite was observed in the three-week data.
increase in resistance was observed for the 17.0lb/gal [2,037-kg/m3] compared to the 16.O-lb/gal
The two conflicting trends
[1,917-kg/m3] system.
suggest that a 3% difference in data is most probably
due to experimental error,
Addition of fly ash to the cement mixture had no
effect on reducing the leaching action of carbonic
acid.
A fly ash/cement admixture containing 35% and
50? fly ash was prepared at 16.O-lb/gal [1,917-kg/m31
density and compared to a neat 16.O-lb/gal [1,917Although the three-week data
kg/m3] cemelt system.
showed improved resistance to carbonic acid leaching
in the 50% fly ash/cement admixture, at tn> end of the
six-week exposure period the percent strength loss in
the fly-ash systems was higher than the neat cement
system (Table 8).
A 66% and 57% strength loss was
recorded for the 35% and 50% fly-ash systems, respectively; whereas, a 51% strength loss was recorded for
the neat cement slurry. Even assuming a (+) or (-) 3%
error in the measurements, the data indicate that fly
ash/cement admixtures are as susceptible to the action
of carbonic acid leaching as a neat cement.
This
finding supports the macrosample data which showed
that a 50% fly ash/cement admixture at 14.O-lb/gal
[l,67E!-kg/m31 and 15.6-lb/gal [1,869- kg/m31 density
had three times as high a carbon dioxide penetration
rate as the corresponding neat cement (Table 1).
Although the addition of fly ash decrease:. the permeability of cement, which should decrease carbonic acid
diffusion into the matrix, a corresponding decrease in
M
an example, a
the cement-to-water ratio occurs.
density
neat cement mixed at 16.O-lb/gal [1,917-kg/m3]
has a cement-to-water ratio of 2.37 by weight; whereas, addition of 35% and 50% fly ash (mixed at the same
density) decreases the ratio to 1.90 and 1.62, respectivelyo
This suggests that a high cement-to-water
ratio is as important as low permeability in controlling the leaching action of carbonic acid in Z;,J
A third factor controlling c~rbon
cement matrix.
dioxide corrosion is the ability to react-up the
calcium hydroxide in the cement matix, thus decreasing its availability to react .. th the carbonic
acid. This is shown in Table 9. BE ause silica sand
or flour does not easily react with calcium hydroxide
below 230F [11OC], addition of 35% silica lowers the
cement-to-water ratio to 1.91 when mixed at 16.0lb/gal [1,917-kg/ms] density without reacting with the
calcium hydroxide in the cement. Compressive strength
loss was on the average 20% to 25% higher than the
corresponding neat cement or fly ash/cement admixture
(Tables 8 and 9).
It is interesting to note that
addition o? a fluid-loss additive has essentially no
The abnormality in the
effect on the end results.
three-week data in Table 9, i.e., the 6% value, at
present is attributed tc a sampling error.
Based on
these data, it was hypothesized that a cement having a
high cement-to-water ratio, low permeability and no
free lime would give the best resistance to carbonic
acid corrosion.

CDF

lb

1K17K
**,

To test this theory, magnesium chloride was used

to prepare a chemical cement admixture.
As reported
by Manmohan and Mehta,6 magnesium chloride, correspending to 2% anhydrous salt by weight of cement,
reduces the pore volume and permeability of cement
pastes. The cement slurries were mixed at 16.0 lb/gal
[1,917 kg/m31 (2.37 cement-to-water rat!o). As shown
in Table 10, addition of 2.5% and 5.0% anhydrous salt
resulted in 24% and 9% strength loss, respectively,
compared to 33% for a O% magnesium chloride system
after a three-week carbonic acid exposure.
Although
an improvement in carbon dioxide resistance was
observed after
three weeks, the s.:-week results
showed a marked increase in strength loss. The 2.5%
system lost 49% of its compressive strength compared
to 51% for the O% system; whereas, the 5% system lost
75% of its strength (Table 10) at the end of the sixweek test period. Because of the limited data and the
marked increase in strength loss after six weeks, it
is difficult to make a definitive conclusion regarding
the proposed theory.
Certainly, the three-week resuits tend to support the theory that a high cementto-water ratio and a low-permeable matrix decrease the
leaching action of carbonic acid in cements.
During the course of this work a serendipitous
discovery (which supports several of the findings in
this study) resulted in the development of a cement
formulation that exhibits a high degree of resistance
to carbcnic acid attack. Compared to the neat Class H
slurry, the XP-1 cement admixtures are on the average
45% more resistant to the action of carbonic acid
leaching (Table 11). At the end of the six-week test,
the Class H system had a 51% strength loss, whereas
the XP-1 formulations on the average .lost 29% of their
An explanation for the imcompressive strengths.
proved resistance is suggested in the three-week
In all cases, the XP-1 systems showed an indata.
crease in compressive strength.
This is due to the
presence of calcium carbonate in the cement matrix.
Because calcium carbonate is cementitious, a resultant
increase in compressive strength occurs; most important, however, the presence of the carbonate indicates
that the diffusion of carbon-dioxide-laden water into
the matrix was decreased.
Subsequently, the equilibrium reaction that changes calcium carbonate to watersoluble calcium bicarbonate is occurring at a slower
If
rate compared to the neat Class H formulation.
this were not the case, a decrease in strength would
be observed.
A third microsample set of experiments was run to
determine the effects of carbonation at lower temperatures.
Carbonation conditions were 125F and 3,000
psi [51.7C and 29.68 MPa].
As was observed in the
175F
[79.4CI
work,
the
3S/65
and 50/50 fly
ash/cement admixtures had less resistance to carbonic
acid attack than the neat Class H cement mixed at the
same density (Table 12).
Strength losses of 58% and
50% were recorded in the 35/65 and 50/50 fly-ash
cements, respectively, compared to a 47% decrease in
Carbonation effects on 65/35 fly
the neat cement.
ash/cement admixture with 2% gel mixed at 14.5-lb/gal
[1,737-kg/m3] density are shown in Table 13 and compared to the data in Table 12.
In the lower density
system, a larger 10S: in compressive strength was
observed (61%), but the degree of strength loss was
not as high as expected based on previous data which
showed that a 15-lb/gal [1,797-kg/m3] density neat
cement underwent a 100% strength loss after carbonation at 175F and 3,000 psi [79.4C and 20.68 MPa]
(Table 7).
This difference was attributed to the

=I!7A

R. A. BRI

SPE 15176

lower permeability that results from the addition of

fly ash to cements.
An XP-1 formulation also was
tested. As shown in Table 14, the XP-1 cement system
had the highest resistance to the action of carbonic
A 26% strength loss was observed,
acid leaching.
compared to 47% in the neat cement, 58% in the 35/65
and 50% in the 50/50 fly ash/cement admixtures.
In
all cases (Tables 12, 13 and 14), decreasing the
carbonation temperature from 175 to 125F [79.4 to
51.7C] did not result in a significant decrease in
the leaching action of carbonic acid on the cement
natrix. Therefore, conclusions made from these experiments are applicable over the carbonation conditions
studied and can be extrapolated to higher temperatures
and pressures.

CONCLUSIONS
The objectives of this study were to determine
the effects a wet carbon dioxide environment had on
portland-base, oil-well cements and to,develop a more
resistant cement to the action of carbonic acid leaching.
Based on the data collected using microsamples,
cements undergo a steady decrease in compressive
strength and show a marked visual deterioration in the
Although a direct
presence of wet carbon dioxide.
correlation between the laboratory data and an actual
wel1, at present, cannot be made, the data suggest
that a cement corrosion problem may develop over the
service life of the well. As cementitious material is
leached from the cement matrix to form an amorphous
silica gel, casing corrosion protection and zonal
isolation will be lost.
Although carbon dioxide
corrosion in portland cements is difficult to stop,
cementing formulations are now available which resist
the leaching action of carbonic acid.
Under the
conditions of most carbon dioxide flood projects,
i.e., low temperature and pressure, the recommended
completion material is the XP-1 formulated systems
mixed at a cement-to-water ratio of 2.3 or greater,
These completion materials are less susceptible to the
leaching action of carbonic acid than the equivalent
density neat cement or fly ash/cement admixture.
Performance data show that XP-1 systems are 45% more
resistant to carbon dioxide attack than either neat
cement, 35/65 or 50/50 fly ash/cement admixtures of
equivalent density.

In addition to the discovery ani development of

the XP-1 completion material, the study clearly demonstrates the advantages of using microsamples over
macrosamples for carbon dioxide corrosion studies. As
shown in the data, the use of microsamples provides a
pseudo-accelerated method to test cements; whereas,
information obtained from 2-in. API cubes and l-in.
diameter cylinders is limited and may lead to erroneous conclusions.
A summary of additional observations made during
the course of this work is outlined below.
o

Portland cements deteriorate in a wet carbon

dioxide environment with loss of compressive
strength and structural integrity.

High cement-to-water
ratio
carbon dioxide resistance.

slurries

Addition of fly ash to cements does not improve

carbon dioxide resistance.

improve

:DORFER

Standard cement additives in low concentration do

not affect the cements resistance to carbonic
acid attack.

Magnesium chloride admixtures improve short-term

but do not improve long-term resistance.

Addition of cement diluents, such as lost-circulation additives and silica (well temperature
below 230F), decreases carbon dioxide resistance.

Initial data indicate that salt does not affect

carbon dioxide resistance in cements.

Extrapolation of the carbon dioxide corrosion

rates in cements observed in this study to a cement
sheath downhole is difficult, but the findings may be
indicative of potential corrosion problems that occur
over the life of the well.
The end result of such
oroblems IS a higher frequency of remedial workovers
and lost production.
In view of the findings of this
study, the r~commended cement compositions along with
good cementing practices is a logical approach to
?nsuring a successful completion in an Enhanced Oil
Recovery project.

FIELD APPLICATIONS
This new cementing formulation (XP-1) was field
tested in the Permian Basin and is now a commercial
pumping service.
To date, about 30 wells were comApplication is usually
pleted with the XP-1 slurry.
across the pay zone plus 200 to 300 ft (61 to 91 m)
above the zone of interest.
General well conditions
were 100 to 115F (38 to 46C) BHST. Well depth was
between 5,000 and 6,000 ft (1,5?4 to 1,829 m).
Completion applications have included liners, as well as
squeeze cementing. Bond logs, as reported by the well
operators, ranged from good to excellent across the
XP-1 application zone.

REFERENCES
1.

Newton, L. E. tlr., Hausler, R. H., eds.: Carbon

!lioxicleCorrosion in Oil and Gas Production,
Selected Paper, Abstracts, and References, National Association of Corrosion Engineers, Texas
(1984).

2.

Onan, 0. D.: Effects of Supercritical Carbon

Dioxide on Hell Cements, SPE 12593 (March 1984).

3.

Biczok,

I.:

Concrete Corrosion-ConcretePrfitec-

tion, Chemical Publishing

(1967) 281-300.

Co.,

Inc., New

York

4.

Lea, F. M.: The Chemistry of Cement and Concrete,

Chemical Publishing Co., Inc., New York (1971)
287-298.

5.

13iczok, I.:

6.

Fianmohan, 0. dnd Piehta, P. I.: influence of

Pozzolanic, Slag, and Chemical Admixture No. 1,
Pore Size Distribution and Permeability of Hardened Cement Pastes, Cement, Concrete, and Aggregates, CCAGOP (Summer 1981) 3, No. 1, 63-67.

Op

tit,

282.

1 s17.6
.

SH
-..

r
S1 HETRICCONVERSION FACTORS
ft

3.048*
(F - 32)/1.8

lb/g;! x 1.198 264

psi x 6.8!2$?57

E-01 = m
= c
E+02 = kg/m 3
E+OO = MPa

TA8LE4
EFFECTS
OF DENSITY
ON PORTLAND
CENEMTSTNENOTki
RETROkiNESSION
INKICROSANPLES
AFTERCARBONDIOXIDE
CAABOMTION
AT 175F
. 3,000P$i [79.4*C, 20.68 !ipa]

*Conversion factor is exact.

Oensity

(lbiqal
) [kghs~

TABLE 1

CARBONATION
OEPTHNEASUREMUiTS+
IN
ONE-INCH
O1#JiETER
CYLINDERS

15

1,797

82

16

1,917

62

S1urryDensity
I
15.5lblgal 17.0Iblgal

14.0lb~
1,678kgid

CementAdmixture

z Strength LOSS

AfterStx-tleek
Carbonation

~857

2,037 kg@

kghx3

0.01 in.

Class H

0.07 In.

0.03 in.

<

ClassH, 50%FlyAsh

0.23in.

0.09in.

< 0.01 in.

31-day carbonation period at 175F, 3,000 psf

[79.4C,

20.68

TABLE 5

MPa],

EFFECT OF 065 ON PORTLANO CENENT STRENOTH RETIWRESSION

IN UICROWPLES
AFTER CARBON DIOXIDE CARBOMTION
AT 175-F. 3,000 psi [79.4C, 20.68 I!Pa]

TA8LE 2

COMPRESSIVE STRENGTH ANO CAR80KATION DEPTH IN

ONE-INCH DMJilTER PORTLAND CEliENTSAMPLES

Compressive Strength
Carbonatedb
psi)
[14Pal

Carbonationb
Oapth (in.~

Camanta
Class A

0.067

6,220

42.88

5,970

41.16

Class C

0.054

6,150

42.39

6,420

44.26

0.083

5,040

34.75

5,820

40.19

Class H

Slurry

density

= 15.6

lb/gal

[1,869

,,

Centrol
[mPal
(P si )

at4ixed

% 065a

X Strength Loss
After Six-Nsak Carbonation

0.0

82

0.5

82

1.0

84

at 15.0 lb/gal

[1,797

kg/m3].

kg/m3].

31-day carbonation period at 175F, 3,000 Psi [79.4C, 20.68 HPa],,

TA8LE 6

EFFECTS ON SALT ON PORTLAND CENENT COMPRESSIVE SIXENGTH RETKKAESSION

IN MCROSAJ4PLES AFTER CARBON OIOXIDE CARBONATION
AT 175*F, 3,0W psi [79.4*C, 20.68 MPa]

TABLE 3
Z Salt (BktOC)a
PORTLAND CEMENT STRENGTH RETRCKRESSION IN RICROSANPLES AFTER CARBON
DIOXIDE CARBONATION AT 175F. 3,000 P$i [79.4C, 20.68 14Pa]

Camxat

X Strength Loss
After Six-Meek
Carbonation

Portland Ca

B2

Portland Ha

80

ai4ixedat 15.0 lb/gal [1,797 kg/m3].

a14ixedat 15.0

X Strength Loss
After Six-Heak
Carbonation

82

18

86

37

82

lblgal[1,707kg/m3].

SISE
TABLE 7

TABLE 10

EFFECT OF DENSITY ON NEAT PORTLANOCENEt@

AFTER CARBON DIOX1OE CARBONATION
AT 175*F, 3,000 PSi [79.4C, 20.68 NPal

EFFECT OF ILAGNESIUN CHLORIOE ON PORTLANO CEIWiTa

STRENGTH LOSS AFTER CARBON OIOXIOE CARBONATION
AT 175F,
3,000
psi [79.4C,
20.68 NPa]

z Strength Loss
6 Weeks
3 Weeks

Density
(lb/gal) [kg/m 31

S Strength
LOSS
3 Weeks
6 }eeks
.

Z Nagnesium Chloride (BkN3C)

100

15

1,797

08

16

1,917

33

51

17

2,037

36

54

aMaryneal

Class

IEs 176

H cement.

aHaryneal

Class

33

::

2.5

24

49

5.0

09

75

H cement mixed at 16.0 lb/gal [1,917 kj/m3] density.

TABLE B
FECTS OF FLY ASH ON PORTLAND CEMENT STRENGTH LOSS AFTER CARBON OIOXIOE
CARBONATION AT 175F, 3,000 Psi [79.4-C, 20.68 MPa]

x Fly Asha

_Source

TABLE

STRENGTH LOSS IN EXPERIMENTAL CEMENT FORMULATIONS

AFTER CARBLM D1OXIDE CARBONAYIONAT
175F,
3,D(N3 psi [79.4C,
20.68 RPa]

Z Strength Loss
3 kfeeks
6 Meeks

..

33

51

35

La Due

36

66

Fonsulationa

50

La Due

24

57

Neat Maryneal H

11

Slurry formulations were mixed with Maryneal Class H cement at 16.0

lb/gal [1,917 kg/m3] density.

aSlurr;,

S Strength
LOSS
3 Ueaks
6 Marks
33

51.

xP-lA

-44

3!

XP-lB

-55

26

XP-lC

-07

29

mixed at

16.0

lb/gal

[1,917

kg/m3]

density.

TABLE 9
EFFECTS OF 35X SILICA
SAND ANO FLOUR ON PORTLAND CEMENT
STRENGTH LOSS AFTER CARBON OIOXIOE CARBONATION
AT 175F,
3,000
psi [79.4C, 20.6B NPa]
TABLE 12
Silicaa

lJ!!!!W

Z Strength Loss
3 Weeks
6 Weeks

33

54

Sand

35

22

70

Flour

35

44

Sandb

35

06

56

Flourb

35

32

76

Slurrymix~d with Maryneal Class H cement at 16.0 lb/gal [1,917 kg/m3]

density.
Slurry contained 0.5 gal/sk 0108 fluid-loss additive.

EFFECTS OF FLY ASH ON PORTLAND CEMENT STRENGTH LOSS

AFTER CARBON OIOXIOE CARBONATION AT
125F,
3,000
Psi [51.79C, 20.6BRPal

% Fly Asha
o

Source

x Strength LOSS
6 Weeks

----

47

35

La Oue

58

50

La Oue

50

aSlurry formulations were mixed with Maryneal Class H cement at

16.0 lb/gal [1,917 kg/m3] density.

TABLE 13

X STRENGTH LOSS INA LIGHTWEIGHT PORTLANOCENENT SLURRY

AFTER ChRBON DIOXIDE CARBONATION AT
125F. 3,000 PS.i [51.7 *C, 20.6B NPd

Dansity
[kgfns]
(lb/gal)

% St~H@~Loss

g Fly Asha

a020
..

16.0

1,917

47

35

16.0

1,917

58

50

16.0

1,917

50

65

14.5

1,737

61

aSlurry
[1,917

formulations
mix[?d with
kg/m3] density.

Maryneal

Class

H cement at

16.0

TABLE 14

STRENGTH LOSS IN EXPERIMENTAL CENENT FWWLATIONS

AFTER CAABON OIOXIOE CARBOFLATIONAT
125F. 3,000 psi [51.7C, 20.68 Wa]

stplln:~Loss

Foraulationa

Neat F4aryneal Class

47

XP-lC

aSlurry

mixed at

16.0

26

lb/gal

[1,917

kg/m3]

density.

lb/gal

COMPRESSIVE

STRENGTH

SLURRY

VS

DENSITY

91

8 *

7 -

-0

CONTROL

Ldn

(yO
l-o
tn o

6 -

EXPERl!wlENTAL

. 5 -

w4 -

3 CIASS
2 -

31DAY

H, 2%

CARBONATION

D79,

PERIOD

1%

AT

175F

3,000

S1

psi

sAMpLEs)

(MACROCYLINDRICAL
~J

o
13

SLURRY

WEIGHT

18

17

16

15

14

(lb/gal)

Figure 1

COMPRESSIVE

STRENGTH

VS

SLURRY

1-

06
z
Wn

&o
(no
l-d -
$x

DENSITY

EXPERIMENTAL

1--o

m
ii
Wm
CEw
n

4 3 -
CLASS H, 50%

2 -

z
o
01

31-DAY

FLY ASH, 2?% D79,

1% S1

CARBONATION PERIOD AT 175F

3,00U
(MACROCYLINDRlCAL

()~
13

sAMpLEs)
I

17

14

SLURR;5WEIGHT
Figure 2

fb/gal)

psi

,~

I
~~