Académique Documents
Professionnel Documents
Culture Documents
SPE 10032
SPE
Society of P@’oleumEngineers
.... ....
“Member . . .SPF.
-, -r
AIMF
. . . ..-
133
. .
to gypsum-cement blends, with or without lightening following as most pertinent and arranged them in a logic&l
agents. Many have used salted cements for Arctic sequence. All cubes referred to were API standard two-
cementing with a common misgiving. that salt melts ice. inch (50.8 mm).
They see in persistent melts around salted slurries poured
into holes in block ice in Iaborator y cold chambers coun- 1. Material Balance
terparts around cement sheaths in permafrost.
We molded a standard test cube from 100 parts API
The API Committee on Standardize Lion4’f Oii-i4ell Class A cement mixed with 46 parts brine containing 30
Cements specifies ~erformance requirements for these parts salt per 100 parts water (by weigt!t). We made a
materials and tests to measure how well cements meet control cube of the same size and w/c (water-cement
them. The committee cent inually refines laboratory weight ratio) from another port ion of the same cement
instruments and procedures so as to impose ever more mixed with distilled water. We removed both cubes from
realistic well conditions in tests. Conditions that most the molds after three days of curing at 140 F (60 C) and
influence cement strength and its rate of development are 3000 psi (20.7 MPa) and immersed each in a separate
temperature and pressure. Specified curing temperatures reservoir of 800 ml of ion-free water. Once a week, later
are currently realistic of well conditions, curing pressures at two-week intervals as changes slowed, we calculated
are not. Until now, pressures have been limited to the material balances on the specimens after analyzing reser-
maximum of 3000 lb (20.7 MPa) due partly to our accep- voir waters and weighing cubes. Reservoir water was
tance of the inaccuracy that higher curing pressures had replaced after each periodic analysis to accelerate
little additional effect on strength, but due mainly to the changes. Results shown in Fig. 2 are for materials that
inherently slow and expensive development of research changed significantly, salt and water; those which
equipment to impose higher preswres. We are now changed little were cieieted. The data were converted
learning better with this equipment, and in due course this intc volumes of material gained and lost for a more
new knowledge will be incorporated into recommended explim i presentation.
testing practices.
2. Cement Cubes and Cylinders
Metcalf and Dresher6 worked at higher pressures
and presented compressive strength data on cement sys- We molded standard cubes and comparably sized
tems cured at temperatures from 170 to 260 F (76.7 to cylinders from a slurry of 100 parts API Class H cement
126.7 C) and at pressures from 3,000 to 10,000 lb (20.7 to mixed with 40 parts brine containing 18 parts salt per 100
68.9 M Pa). They reported pertinent data for an API parts water. After three days curing at 140 F (60 C) and
Class H cement slurry containing 18 percent sodium 3000 psi (20.7 MPa) we removed the specimens from the
chloride. These showed the 24-hour compressive strength molds, immer%ed them in tap water and aged them at 75 F
at 170 F (76.7 C) for this system dropped from nearly (22~8$). Fig. 3 is a photograph of the specimens after six
6000 to less than 3000 psi (41.4 to 20.7 MPa) when the .
curing pressure was increased from 3000 to 8000 lb (20.7
to 55.2 MPa). Metcalf and Dresher reported that x-ray 3. Shale-Cement Cube Halves
diffraction analyses fz%?d to disclose the cause of
strength losses. R k, p~ssible tk “i ‘“-* fIShIg pressure We poured slurries of API Class A cement mixed
increases the amount of chloride ion which combines with either with fresh or salt saturated water around pieces of
cement as Ludwig described, and that the increased water+ ensit ive Miocene shale in standard cube molds,
combination fu ther weakens cement. covered and immersed them in a water bath at 120 F (48.9
C) for 48 hours. We removed the hardened cubes from the
The user must judge for himself whether half or full molds and cut them into halves exposing internal faces
saturation is sufficient or necessary for properly cement- each consisting of shale surrounded at least on three sides
ing through salt. Cement bond logs, communication tests by cement. We immersed the halves in waters that were
and other evaluations can help him decide. There are either fresh or that contained varied amounts of salt and
reasonable expectations from references cited and from aged them eight months at atmospheric pressure and 75 F
theory that the other uses of highly salted cement for (23.9 C). We removed, dried, stained with kerosene for
inhibiting hydration and depressing slurry freezing tem- better contrast and photographed the specimens which are
peratures will meet those objectives. What has been pictured in Figs. 4, 5, 6, and 7.
lacking, except for the Metcalf and Dresher data, is
information on what can happen to the cement itself as a We pretested the shale used in this experiment for
result of salting. The purpose of this paper is to provide water sensitivity by immersing pieces in fresh water. The
it. Data and pictures are presented which suggest the pieces first became fractured, then crumbled and finally
continuing decline in strength with time. These further disintegrated after a few days.
show that deterioration is more rapid and failure more
certain if salted cement remains in contact with less EXPERIMENTAL FREEZE-THAW CYCLING BEHAVIOR 01
salted water. HYDRATED CEMENTS IN WATERY ENVIRONMENTS
EXPERIMENTAL REACTIVITY OF SALTED CEMENT The freeze-thaw tests in the following experiments
foliowed the alternate method described in the last para-
We (authw and co-workers) conducted many experi- graph of lb API Arctic Cementing Testing Procedure
ments on salt cement deteriorate ion during the past ten reproduced in the Appendix. The regular procedure
years to find its cause and prevention. We selected the decelerates changes in that specimens are cycled in molds
with single cube faces exposed; whereas, the alternate eight weeks. Osmosis effected the exchange. The
procedure specifies removal from molds and exposure of cement behaved initially as a series of semipermeable,
all six faces. Changes occur with the third or last cycle then permeable, membranes of silica with microscopic
in the alternate which never appear in the regular proce- capillaries that allowed the influx of water and efflu of
dure. salt. The volumes of wat ~ and salt exchanged, 19 cmy, is
14 percent of the 131 cm volume of the cube. Osmotic
4. Salted Gypsum-Cement pressure created high stresses in the specimen during this
relatively large exchange. We began to see the effects of
The specimen shown in Fig. 8 was originally a cube these stresses as we watched the specimens longer.
of a typical Arctic cement consisting of 50-50 gypsum-
cement mixed with water containing enough salt (18 Aging was prolonged in Exp. 2 (Fig. 3) on cubes and
percent) to lower the freezing point to 15 F (-9.4 C). cylinders to observe the final consequences of stresses.
All cubes and cylinders in Fig. 3 display multiple fractures
5. Low Density Cement - A after six months of aging. It was necessary to tape the
specimens together to photograph them.
The specimen shown in Fig. 9 had been a standard
cube made from a slurry containing the following mate- In Exp. 3 fresh water cement kept the shale intact
rials in parts by weight: although the specimens were aged in fresh water (Fig. 4).
There are no signs of weakened cement or shale nor
100 API Class B cement damage to the bond between them. The same description
30 Gilsonite applies to the fresh water and salt+ aturated shale-
1 calcium chloride cement halves aged in saturated salt water (Fig. 5). Both
59 water. sets of specimens were nearly in equilibrium with their
aging waters. Little stress developed in the fresh water
6. Low Density Cement - B specimen aged in saturated salt water (Fig. 5) as there
was little free water in the cement. Of most significance
The specimen shown in Fig. 10 is a cube made from among aged specimens was the inhibition of this shale by
a slurry containing the following materials in parts by a cement slurry that contained only a few hundred parts
weighti per million of calcium ion. Salt supplements were un-
necessary.
100 API Class B cement
30 Gilsonite The severity of fracturing in the specimens shown in
3 bentonite Figs. 6 and 7 is proportional to differences between
1 calcium chloride original salt concentrations in the cement and aging
L3 water. water. Developments from Exp. 2 provide the basis for
concluding that fractures began in the cement and con-
DISCUSSION OF SHALE INHIBITION tinued into the shale. Furthermore, the shale appears
unaffected by cement or aging water in any specimen.
Salting can weaken cement in three ways: by There would have been no damage to any specimen if the
making a noncementitious dilution, by combining with ion concentration in the aging water had been the same as
silicates, by inducing physical reactivity. Dilution and in the cement.
combination are the only ones that weaken cement adja-
cent to salt. Half saturation instead of full reduces When we drill a well with a Iow+mlt mud but set
dilution losses. Usually, combination loss should be mini- casing with a high-salt cement, we probably inhibit shale
mal as Metcalf and Dresher found it proportional to but we certainly damage cement. All the specimens from
confining pressure, and as most massive salt bodies Exp. 1, 2 and 3 display the same consequence of confining
cemented are relatively shallow and low pressured. hardened cement with Iess+alted water. There have been
Except in a rare instance, neither saturation nor half no exceptions, including specimens not pictured. As
saturation causes strength loss sufficient to deter the user osmotic pressure is independent of confining pressure, it
from his choice of concentrations to cement salt forma- is unlikely that aging specimens under a few thousand
t ions. But in uses to inhibit shale, salting initiates a pounds of pressure would have changed results much.
sequence of changes that can disintegrate cement.
Never salt ing cement is not the solution to the
Primary cement sheaths commonly overlap different inhibition problem. Salting mud with sodium or potassium
kinds of formations. It is tedious and impractical to chloride is now recognized as essential to drilling and
displace incremental slurries so that only specific ones stabilizing shales and sands sensitive to fresh water.
harden against certain formations. Primary cementing Salted filtrate from these muds invades fissures and
Usl]ally is a compromise in tailoring a slurry for priority fractures in the shale. To the detriment of cement and
zones with lesser concern for secondary. Part of a slurry shale, an unsalted slurry against this shale would lose its
salted mainly to inhibit shales wiU frequently harden also water to dilute the salt in the fissures and fractures.
against a fresher-water sand. Experiments 1 and 2 Balancing ions in the cement with those in the formation
examined this consequence. is the solution indicated for all these problems, but it is a
difficult solution to apply.
Material balances plotted in Fig. 2 (Exp. 1) reveal
that water replaced virtually all salt in the test cube in
* w....”-~v- ----- -. . . . . . . . . - ..-— — --..-—- ..- -- —-- ----
DISCUSSION OF FREEZING POINT DEPRESSION same osmotic exchange illustrated in Fig. 2 proceeded in
the gypsum-cement specimen until influx water had
Wolfgang Czernin7 examined freeze probabilityies= diluted the selt enough to have a freezing temperature
water existing in three possible states in hardened con- above 20 F.
crete. He describes the chemically combined water as
‘non-f reezable’ as is also the gel water which freezes at No exihange occurred in Experiment 5. The rubble
very low temperatures owing to the surface forces to shown in Fig. 9 resulted from the freeze of the 19 percent
which it is subjected. Thus, under normal climatic water in excess of the 40 percent limitation previously
conditions, it is only the capillary water which can be described and tested. The specimen froze in the first
regarded as freezable. But even the capillary water does cycle as there was no time lag required for inbibing free
not freeze completely as soon as the temperature drops to water.
the freezing point of ‘water. This is due to the fact that
capillary water is not pure water but an aqueous solution Priority concern for freeze in the slurry is mis-
of a number of salts, mainly of calcium hydroxide and placed if it ignores the more imminent freeze in the
alkalis. Its freezing point should correspond roughly to 30 hardened cement, especially when salting to prevent the
F (-1 C). In considering the porosity conditions in former virtually assures the latter. There are meesures
hardened cement paste, Czernin maintains that no capil- other than salting that can be taken to delay I ~” eze
lary water remains after complete hydration if the w/c beyond the time necessary for an accelerated set. A
ratio of the paste is 0.4 or less. The assumption that only warmed slurry supplemented by drilling fluid warmed
the capillary water can be considered as practically casing comprise a sensible heat bank that would take a
freezable would indicate that a frost-resistant concrete long time to exhaust to the slurry freeze, perhaps days.
must be obtained if the w/c ratio used in its preparation Experiment 6 revealed another measure that if taken will
lies below about 0.4. remove the danger of freeze in the hardened cement.
Ease in pouring and working concrete into surface The contrast between specimen appearances in Figs.
forms requires adding little more water than necessary 9 and 10 is due entirely to the inclusion of three percent
for fully hydrating the cement. Limiting water conserves smectite (bentonite) by weight of cement when the sur-
ultimate strength and reduces concern for post-hydration viving cube was prepared. Otherwise, the compositions
freeze if the Czernin threshold of 0.4 is valid and not were identical. Thecompressive strength of the cube was
exceeded. There is more concern for cement slurries more than 2000 psi (13.78 MPa) at the time it was
pumped around casings in permanently frozen formations photographed. The same addition of bentonite also pro-
where slurry mobility requires water fractions more than tected other arctic type cements including those salted
twice those of concrete. from freeze damage during cycling.
During this study we explored water-cement ratios One explanation for the protection afforded by
around 0.40 for freeze in hmdened cements. We found bentonite involves comparisons of forces binding water to
that 0.40 is a critical value allowing no more than two or clay ~d forces of ice crystallization. Gregor and
three hundredths in excess water. Hardened cement Gregor recently suggested a more novel one. They wrote
cubes with wlc!s of 0.44 to 0.50 dkintegrated on thawing of two researchers, J. D. Bernal and R. h. Fowler, who in
after the second, sometimes the first, freeze to 20 F (-6.7 1933 demonstrated that inclusion of a very small amount
C). Similar specimens with w/c’s of 0.40 and below were of hydrophilic material caused water frozen in a semi-
undamaged after several freeze-thaw cycles as evidenced permeable membrane to be in an ice lattice instead r“ ‘
by increased compressive strength and cube integrity. crystal. The lattice is distinguished from the crystal
its discontinuity and relatively few molecules per un:‘i.
Freeze-thaw cycling simulates conditions impe:ed Freeze expansion is apparently absorbed in the spaces
on cement around casing in permafrost for a well alter- between lattice units.
nately produced and closed in. Permafrost is an all-
inclusive name for frozen formations ranging in composi- CONCLUSIONS
tion from all-water to silt, sand, and/or gravel mixtures
devoid of water. Nearly all slurry water is freezable 1. Cements salted to half saturation or more which
before hydration; according to Czernin only capillary harden against formations other than massive salt
water is freezable afterwards. The common method of likely deteriorate with time and to an extent pro-
preventing freeze consisting in adding an electrolyte, portional to depth and confining pressure.
usually 18 percent sodium chloride, depresses the slurry
ireezing temperature to about 15 F (-9.4 C). This works 2. Optimistic initial com@etions through salted
in fluid slurries and continues working in hydrated cement cement may be misleading as deterioration of
isolated from water, but protection breaks down above 20 cement sheaths may not become evident for many
F in hardened cement contacting alternately thawed and months, perhaps years.
frozen water of lower salinity. This accounts for the
damage to the specimen described in Experiment 4 and 3. The only way to anticipate halting salted cement
pictured in Fig. 8. deterioration short of destruction is to balance ion
concentrations in the cement slurries and forma-
Rarely does permafrost water contain more than tions.
three percent salt and only then when its source was sea
water. So the ion differential that preceded the specimen
destruction shown in Fig. 8 was of the magnitude to be
expected in the thawed medium of true permafrost. The
4. Salting cements to depre.s slurry freezing tempera- 3. Slagle, K. A., and Smith, D. K.: “Salt Cement for
tures is unnecessary and self defeating. Accelerat- Shale and Bentonitic Sands”, J. Pet. Tech. (Feb.
ing and warming slurries preclude freeze before ?.963) 187-194; Trans. AIME, 228.
initial set; inclusion of a hydrophilic material in the
slurry formulation prevents freeze damage after- 4. “API Specif icat ion for Oil-Well Cements and
wards. Cement Additives”, API Spec lOA, 20th cd., API
Prod. Dept., Dallas (1979).
ACKNOWLEDGMENT
5. I?API Recomme!lded Practice for Testing Oil-kfeu
I express my appreciation to the management of Cements and Cement Additives”, API RP 10B, 21st
Gulf Science and Technology Company and Gulf Research cd., API Prod. Dept., Dallas (1979).
& Development Co., for encouragement to prepare this
paper and permission to publish it. This appreciation 6. Metcalf, Arthur S., and Dresher, Thomas D: “The
extends to co-workers, particularly W. H. Grantt Jr., Effects of Pressure on the Set Properties of
whose patient and precise work made this paper possible. Cements with Various Additives”, paper SPE 6800
presented at SPE 52nd Annual Fall Meeting, Denver,
REFERENCES Oct. 9-12, 1977.
1. Ludwig, N. C.: ffchemistry of Portland Cement 7. Czernin, Wolfgang Cement Chemistry and Physics
Used in Oil Wells”, Oil-Well Cementing Practices in for Civil Engineers, Chemical Pub. Co., New York
the United States, API, New York (1959) 27-33. ?1962).
2. Ludwig, N. C: !!Effects of Sodium Chloride On 8. Gregor, Harry P., and Gregor, Charles D.: l’Synthe-
Setting Properties of Oil-Well Cements”, Drill. and tic - Membrane Technology”, Scientific American
Prod. Prac., API (1951) 20-27. (JuIY, 1978) 112-128.
CONSEQUENCES OF SALTING WELL CEMENTS SPE 10932
APPENDIX
*SECTION 17
ARCTIC CEMENTINGTESTINGPROCEDURE
I * Reproduced from API RP 10B, 21st Ed., Dec. 1979, by permission of American
Depart ment.
Petroleum Institute, Production
136
mm SODIUM cHLoRIDE
—\
m -4 —N,
4 :* z ‘~,
u .-
w --N TEST
~ -8 -- --- ---- f ------ -
s -,0
r 1
tL~ 9 4 6 s 10 12
WEEKS
.;;57
., ,#-,
‘“y
.“
.
,.’.
,-, ...,.
b..
. +-AL,.,
,’<i
->:=:~:,r :.
,..-,, -.~y LA.
..~;
‘
. .
:“
,,
,.,
Fig. 4- Fresh water cement aged in Fig. 5- Fresh and saturated salt
fresh water. cements aged in saturated salt water.
.
m
‘!!!!!!
Fig. 6- Saturated salt cement aged in
100/0 salt water.
‘4, ~a -%
~~#dJlll@~
Fig. 7- Saturated salt cement aged i n Fig. 8- Salted gypsum cement after two
30/0 salt water. freeze-thaw cycles.
141
,
142