Cement Bonding

Characteristics
in Gas Wells
Roy S. Marlow, SPE, Southwest Research Inst.

Summary. A program was per- Introduction and Background (i.e., the percolation of gas into the cement
formed to study the effects of cyclic Gas intrusion into cemented wellbores and during setting and curing and the resultant
the resultant leakage to either the surface or creation of leakage paths) and methods
pressure/temperature fluctuations of designed to prevent gas migration at the
porous formations below the wellhead have
gas storage wells on annular leak- been persistent problems in the gas indus- cement/casing and cement/formation inter-
age. A review of pertinent literature try for many years. These problems have faces. Some researchers categorized these
was conducted, and operators of gas resulted in significant safety hazards and ec- as two separate failure mechanisms, while
onomic loss to operators. others categorized them as separate cases of
storage wells were surveyed to de- Through the efforts of a number of a single mechanism. In either case, it is evi-
termine their experience in the area. researchers in the past 2 decades, cement- dent that means taken to prevent the forma-
Results indicated that for most of the ing techniques in general have been vastly tion of leakage paths through the cement
improved and have helped reduce the mag- itself will also help prevent the formation
gas storage wells currently in opera-
nitude of the problem. Little work has been ofleakage paths at the interfaces. Additional
tion, surface leakage is not a prob- done, however, to determine the effects of preventive measures may be necessary,
lem. Where leakage is reported, pressure/temperature cycling on the bond- however, to minimize or to prevent leakage
cyclic fluctuations do not appear to ing characteristics of annular cement to at the interfaces.
casing. Data of this type are of particular Most research has concentrated on the
be a significant cause. In the majori- importance to operators of gas storage wells prevention of leakage through the cement
ty of cases, leakage occurs within the because these wells operate on periodic in- column itself, and one fairly well developed
first few cycles, indicating that it is jection/withdrawal cycles with associated theory explains the need for prevention of
pressure/temperature fluctuations. these types of failures. This theory states that
caused more by static than by dy- The current project was initiated in an ef- annular gas flow is caused by a hydrostatic
namic loads. The only variables fort to begin study of this problem. It was pressure loss sometime between placement
found to correlate with leakage at a conducted in two phases. First, available of the slurry in the wellbore and the devel-
literature was reviewed to identify previous opment of sufficient static gel strength to
high level of significance were depth
work in cementing technology in general and resist the percolation of gas into it. 24 Stat-
and bottomhole pressure (BHP). in the area of annular leakage of oil and gas ic gel strength is that internally developed
Wells that leaked tended to be deep- wells in particular. Then, operators of gas rigidity in the slurry that resists a force
er and had higher pressures than storage wells were surveyed to determine placed on it. The gelling process will begin
the magnitude of the annular leakage prob- immediately after pumping has stopped and
those that did not. The only way to lem, to identify similarities or differences will continue until the cement develops a set.
stop leakage effectively at these in wells with known annular leakage, and As gel strength increases, the cement
higher pressures is the use of me- to establish typical environmental/operating column begins partially to support itself and
parameters of gas storage wells. its volume decreases owing to loss of filtrate
chanical sealing mechanisms.
to the formations and/or to hydration.
Literature Review Hydrostatic pressure caused by the slurry
A computerized search of the open litera- column lessens as the column begins in-
ture was conducted by use of the COMPEN- creasingly to support itself and as volume
DEX data base (Dialog Information Services changes occur. If volume changes are large
Inc.). From this search, 24 references were enough and if the gel strength is not yet suffi-
selected for more detailed review. These cient to resist gas intrusion, leakage to other
works were thought to describe fairly well producing zones or to the surface (directly
the general evolution of thought from the or through a "leap-frog" approach from
early 1960's until the present on the causes zone to zone) may occur.
and means of prevention of gas leakage (also Research carried out in the 1960's and
called gas migration and annular gas flow) early 1970's dealt with the more specific
in cemented wellbores. 1-24 A more detailed case ofleakage at the cement/casing and ce-
review of the papers than given here can be ment/formation interfaces. Carter and
found in Ref. 25. Evans,2 working on the effects of different
Past research efforts can be divided into surface coatings on pipe, showed that static
two broad categories: methods and materi- bond strengths (leakage pressures) using gas
als designed to minimize the leakage of for- ranged from 15 to 400 psig [0.1 to 2.76
mation pressures through the cement itself MPa] for different pipe surfaces. They used
Copyright 1989 Society of Petroleum Engineers test models that were cemented and cured
1146 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - November 1989 • JPT
 1.1 3.0
1.4 2.1
U 2.1

1.2 2.4

1.1 2.2 30._

1.0 2.0

0.' i1 1.1

II
I.'
it
Dol

0.7 1.4

0.' 1.2
:>
z 1.0
0.1
D.4 o.a •
U 0.'

D.2 0.4

0.1 0.2
0.0
D."" 0.1lIIll 0.11
0.0
I 2 3 4 I • 7 • • 10 It 12 13 14 .. .. t7 .. .. 2G 1-1 1-2 2-3 J-4 4-1 5-1 1-7 7-1 1-1 .-,0
_nllI HOl.£ llEPIH (1000 FT.)

Fig. 1-0perator-reported wells (6,953 total). Fig. 4-Reported hole depths (6,901 wells).

3.1
'.0

3.0
1.0

u
4,0

I]
2.D

1.1

1.0
II 3.0

2.0'

o.a 1.0

4,1lIIll
0.0
<1 • • ""-70 1171-7. 1171-10
''''-11 0.0
OPEN HOI£ UNEIl GUN PERP'OMlED
~ _.-uD (6.062) llIII"!!!Jl'I::la IIITH..- (4281

Fig. 2-Performance by period (6,062 wells). Fig. 5-Completlon types (6,860 wells).

4.0 1.0....-----------------------,
3.1
5.0 ....211

3.0

4.0

h 2.0

U
:>
z
2.0

1.1

1.0
II
z
3.0

2.0

1.0
0.1
2._
0._
0.0
7...._
3.1-4.1 5-1.121 1.125-10.71 >10.71 WULnpL£

Fig. 3-Casing sizes (6,722 wells). Fig. 6-Zone types (5,480 wells).

at ambient conditions. Bearden et al., 3 in field only by using mechanical sealing largely agree with Bearden et al.'s for the
work shortly following this, used models elements. zero-pressure-differential case.
that were cured under various combinations In 1967, Pittsburgh Testing Laborator- Later laboratory and field investigations
of internal casing and annular pressures. ies 26 (PTL) conducted a similar series of concentrated mainly on improvement of
They reported bond strengths of 200 to tests using models prepared with very close- general cementing technology and looked at
3,100 psig [1.4 to 21.4 MPa], with the best ly simulated field conditions. They reported the effects of mud density, setting time, tem-
case being one of high annular pressure and bond strengths of < 500 psig [ < 3.45 MPa] perature, gelation, dehydration, bridging,
borehole mud removal, pipe movement and
ambient internal pressure (maximum pres- on models without mechanical sealing ele-
centralization, and the prevention of chan-
sure differential across the casing wall). ments and a 5,OOO-psig [34.5-MPa] limit neling the cement column. Relatively little
They also reported a general leveling off of pressure (without leakage) with mechanical research concentrated specifically on leak-
bond strength with increasing annular pres- sealing elements. PTL pressurized both the age at the annular interfaces.
sure and were able to seal higher annular inside and outside of the casing during cur- Tests show that even when the most up-
pressures both in the laboratory and in the ing, which resulted in bond strengths that to-date cement types and techniques
JPT • November 1989 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 1 l 4 7
 LO s.a

7.0 3.0

'.0 ... 7L....

La
h "0

II
4.0

3.0
i!
z~
I.e

l.O
2.0

1.0

0.0
.. -
IlIMC11ClIW.
0.'

0.0
0-26 20-00 00-70 150-171
SE111NO 1IlIE (ItI.)
Cl.21l1

200-220
D.'..
111-720

Fig. 7-Hole type (6,905 wells). Fig. 10-Cement setting time (3,297 wells).

... 3.0
2JI
"4
2.1
31._
2.2
2.4
PZ ,. POZMIX
..0 2.2
LAT "" LATEX
I.' ~
..0

ij
A,C,G,H Standard API Class Cements
I.' "/" - COmbination of Classes Shown I.e

1.4 I••

II
t.2

'.0
o.a
it
Z
l.4
1.2
1.0
o.a
0.' 0.'
0.4

0.2

0.0 .-
e/H
.31l1

lATtn P'l/A 1/21'% P'l/O 1'% anD

0.4
0.2
0.0
.2O-.e .5-.715 .75-1 1-1.21 1.2&-1.5
--.u5l11Z£(IN.)
1.5-1.71 1.75-2
0._
2-2.21

Fig. a-Cement types (5,689 wells). Fig. 11-Annulus size (5,308 wells).

4.0 ...
... 12.'"
2.4

2.2

3.0 2.0

... ii
u
u
1.4

U
2.0
t.2

II I.a ::>
z
1.0

o.a
'.0
0.'
0.4
0.'
'.27. 0.2
0..... 0.11'1 0."" 0.11.
0.0 0.0
0-. 1-10 20-2' 00-10 11-12 12-1J 13-'4 14-'1 15-1'
COIENT WEJOHT (LB./""'-)

Fig. 9-Cement placement times (3,573 wells). Fig. 12-Cement weight (3,941 wells).

are used, leakage can and will occur in a sig- complete knowledge of all parameters faces. These paths generally are minimized
nificant number of cases. Watters and affecting it. These include temperature, by increasing the bond strength to the sur-
Sabins, 15 for example, reported that, in a pressure, and caliper profiles of the hole, faces. For the cement/casing interface, this
study of 250 casing jobs over a IS-month rheological information on the cement,
means coating or otherwise altering the
period with new compressible cements, 15 % equipment limitations, design analyses of the
of the wells leaked. casing string, and information on the tech- original casing surface. For the cement/for-
As the causes of the annular leakage prob- nical and economic aspects of similar jobs mation interface, this means adequate
lem have slowly become known, so have the in the past. After the job, acoustic or simi- removal of mud filter cake through the use
methods for its prevention. These general- lar types of logs should be run to help en- of chemical washes, scratchers, and casing
ly fall into four categories. sure that no leakage occurs. rotation/reciprocation. Mud removal also
Cementing Program Design and Quality Minimizing Leakage Paths at Cement/ will eliminate the creation of channel leak-
Control. The job should be designed with Casing and Cement/Formation Inter- age paths in the slurry by contamination.
1148 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - November 1989 • JPT
 I .• . . , . - - - - - - - - - - - - - - - - - - - - - - - - , ...
...
1.11
1.4
...01.
I.'
I.' 3.0
17.41.
1.1
1.0 2.1

II
0••

0.' 2.0

•
0.7

0.'
0.'
iI 1.'

0.' ,.0
0.'
0.' 0.'
0.1 2.3. . 1.1. 1.-
0-.5 .1-1
1-' .-. .-4 ....
0.0 ..L.L~:..£.L......l~T.£..~..L..<::...<j.L£.j'--~"i""---LL.4.L.L..l..-L..L4'~ 0.0
0-41.' D.1-1 1-1.1

-(lCIIlI)
1.1-' 2-2.1 2.....

Fig. i3-Uncemented annulus (2,267 wells). Fig. i6-Minimum well pressure (4,369 wells).

..0 . , - - - - - - - - - - - - - - - - - - - - - , LO..,.------------------------,
2.1
2.1

2.' '.0
2.2
'.0

~
I.' :r.o
I .•

I.'
I.' 2.0

1.0
0••

0.' 1.0

0.'
0.' 0..... O.22ll
..... 0.'" ,'- 0.31.
0.0 .L..t.,..:..u--'CL.<TL.<.J....--'-''-Lr£.LL-LL~.LL.-__,---_r_
...... 0.0
0-1 1-2 2-3 3-4 4-5 >1 10-71 71-100 1DO-126 121-110 110-17.
_ _ (IIAX.- ..... KIlO)
lDPIMTIN(lIU.1')

Fig. i4-Pressure range (5,425 wells). Fig. 17-Number of cycles to detect leaks (421 wells).

1.0..-----------------------, 3.0

2.1 2.1
2.1

2.' ...
2.1

2.2
2.2
2.0 31.... 2.0

1.1 1.1

II 1.1

1.0
II ...1.'
1.'
1.2
,.0

...
0••
0.'

0.' 0.'
0.2 0.2
O.22ll
0.0 .LI.~'-£J---J~~.u.......L..<:.4£.LL-L..£.~.LL.--L~.L.L.-'--_r_ ...... 0.0
0-1 1-' 2-' "'" 11-10 1!IO-71 71-100 100-121 12&-110 110-171 17"'200
-(lCIIlI)

Fig. is-Maximum well pressure (5,425 wells). Fig. is-Maximum temperature (5,371 wells).

Cement Density, Placement, and Curing. overcome temperature problems in a hole of the column during setting, are highly ef-
It is imperative that the slurry density be that might cause undesired setting charac- fective.
sufficient to contain the formation pressure teristics and/or to help ensure the shortest For all cement types, the cementing pro-
and that it be placed efficiently to eliminate possible setting time. Special tail-out ce- gram design should be based on the static-
voids. The use of the highest possible pres- ments to ensure that setting occurs from the gel-strength properties of the cement.
sure differential across the casing is desir- bottom of the hole up are commonplace. Ce-
able, as is the fastest displacement of the ments with fluid-Ioss-control additives are Industry Survey
cement and the longest possible curing time. available and are highly desirable. Expand- To help quantify the annular leakage prob-
ing and impermeable cements may have ad- lem in gas storage wells, a survey was pre-
Use· of Proper Cement Type. Cements are vantages in certain cases. Finally, the new pared and sent to the members of the
available for a wide range offield situations. generation of compressible cements, de- American Gas Assn.'s Pipeline Research
Inhibiting/retarding additives can be used to signed to help maintain the hydrostatic head and Storage Reservoir Supervisory commit-
JPT • November 1989 ------------------~-------------------1l49
 3.2 100.0
:s.o 10.0
u 10.7.
u 10.0
2.4
70.0
2.2
2.0
lA
i
~
10.0

00.0
1.1
,...
II 1.2
1.0
I .....,
30.0
o.a

-
20.0
D.I
0,4 10.0
.
...
0.' 0.7.
0.0
0-1 1-3 1-1. >1 •
0.0
-.I.-I ..... 21-100
LfAIlNlE (1000 . ,/OItf)
100-210 >210

I
Fig. 19-Pressure-cycle times (3,669 wells). Fig. 22-Estlmated leakage (135 wells with leakage).

ooo....----------------------,
700
-- 71.7.

i -
100 300

~ 200

I 110

100

.....
10

0-1 0-1 1-3 3-20

-
...... -.HI 10 IIEIB:r LfAIlNlE
....,10 "0-310 >310

Fig. 20-Temperature-cycle times (1,144 wells). Fig. 23-Number of months to detect leaks (426 wells).

7.0 100 ~-------------------___,

f.O H.l711
400

a.o
300
4.0

II 3.0
200

2.G

'00
1.0

,.... 0." 0.07ll 0.1" D.'" 0.""" 2.13X

0.0
ZIIIO -.I.-I 1-21 21-100 100-210 >210 0-3 3-10 10-50 50-100 100-200
NUJ.lBER CYCLES TO DETECT L.£AKAGE

Fig. 21-Estlmated leakage (5,032 wells reported). Fig. 24-Number of cycles to detect leaks (421 wells).

tees. The survey attempted to determine the nology in the past 2 decades have resulted Twenty companies responded to the sur-
magnitude of the annular leakage problem, in improved leakage performance. vey with 30 individual survey forms. These
to establish representative physical and en- Respondents were encouraged to respond covered a total of 6,953 wells. Frequency
vironmental conditions, and to help deter- with "average" or "representative" figures histograms showing the reduced data are
mine similarities and differences in wells shown in Figs. 1 through 24. The signifi-
covering groups of wells or fields and to use
with and without annular leakage. cant number of "gaps and voids" in the data
their judgment in determining which group-
Questions were limited to subjects iden- result from differences in the completeness
ings represented significant differences in of data bases for individual responding com-
tified from both primary and secondary data
sources as highly significant to the problem. history and observed performance. They panies. Histograms were plotted for wells
Respondents were asked to break their an- were also asked to identify items not cov- where all information was available for that
swers into five time periods to see whether ered in the survey that they felt were im- histogram only. The number of wells in-
the general improvements in cementing tech- portant. cluded in the histograms is given in the

1150 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - November 1989 • JPT

figure captions. In Fig. 22, for example, were simply added. These data are shown
leakage rates were reported for only 135 of in Table 1.
the 428 wells with leakage, so only these Table 2 gives the most frequently given
could be included in the graph. Similarly, responses to the categories shown. This ta-
in Figs. 23 and 24, operators reported the ble does not represent a "typical" well; the
number of months to detect leakage for 426 wide variance in well parameters and ce-
ofthe 428 wells that leaked and the number menting/completion practices makes it very
of cycles to detect leakage in 421 of the 428 difficult to determine typical conditions.
cases. Some of the more important variables
This approach allows the maximum were also analyzed for statistical signifi- "Tests show that even
amount of survey information to be present- cance: time period that the well was drilled; when the most up-to-
ed. It may be misleading, however, to try casing size; annulus size; depth; cement date cement types and
to compare some seemingly complementary type; cement setting time; uncemented
graphs. Such comparisons, and certain cal- length; completion type; hole type (straight
techniques are used,
culations, can be made from the graphs, but or directional); zone type (single or multi- leakage can and will
these should be done with caution and an ple); maximum and minimum pressures; and occur in a significant
understanding of how the graphs were con- maximum and minimum temperatures. Data
structed. were analyzed with nonparametric sign-and-
number of cases."
Several of the survey questions asked for run tests. Although easier to apply, these
a "yes" response if certain procedures were methods may have a lower power efficien-
typically used in the field but did not ask for cy (ability to differentiate between statisti-
an estimate of the number of times used. Nor cally significant data) than the more rigorous
was the respondent asked to differentiate be- parametric methods. A comprehensive anal-
tween wells that did not use these procedures ysis with the parametric methods was out-
and wells where the information was not side the scope of this project. Note,
known. For these items, positive responses however, that the more rigorous approach

TABLE 1-RESPONDENT CEMENTING DATA TABLE 3-RESPONDENT·ASSESSED REASONS

FOR LEAKAGE AND REMEDIAL ACTIONS
Technique Number of Wells
Chemical wash after cementing 1,700
Casing reciprocated and/or rotated 1,376 Major causes of leakage *
Stage collars used 315 Microchanneling in cement (5)
Noise log run after cementing 974 Poor cementing job (4)
Permeable zones encountered in Failed wellhead seals (3)
cemented annulus 5,988 Casing-thread/collar leakage (2)
Compressible cement used 655 Charged shallow zones (2)
Expandable cement used 639 Poor casing/cement bond (2)
Separate tail-out cements used 1,836 Collar leaks above cement top (2)
Fluid-loss-control additives in cement 2,314 External casing packers (2)
Accelerators/retarders used in cement 4,330 Casing shoe leaks (2)
Centralizers/scratchers used 4,552 Permeable thief zones (1)
Miscellaneous completion problems (1)
Sloughing shale zone (1)
TABLE 2-REPRESENTATIVE GAS Differential coefficients of expansion between steel and
STORAGE WELL CONDITIONS cement (1)
Loose connections (1)
When drilled Before 1965 Corrosion (1)
Depth, ft [m] 1,000 to 2,000 [305 to 610] Injection/withdrawal pressure changes (1)
Casing size, in. [mm] 5112 or 6 5/a [139.7 or 168.3] Swamp gas (1)
Annulus size, in. [mm] 11/2 [38.1] Coal gas (1)
Hole type Straight Improper mud removal before cementing (1)
Completion type Openhole
Number of producing zones one
Cement type "A" How annular leakage was controlled
Cement weight, Ibm/gal [kg/m 3] 14 to 15 [0.168 to 0.180] Perforate and squeeze cementing (8)
Special additives . No Run liner (4)
Uncemented annulus, ft [m] 500 to 1,000 [152 to 305] Repack wellheads (3)
Cement placement time, hours 4 Monitor pressures (2)
Cement setting time, hours 24 Vent and/or blowdown (2)
Chemical wash used No Well logging (1)
Centralizers and/or scratchers used Yes Not controlled (1)
Casing reciprocated or rotated during cementing No Collars/packers (1)
Noise log run after cementing No Remove and replace bad connections (1)
Maximum hole temperature, OF [0C] 100 to 125 [38 to 52] Plug well (1)
Minimum hole temperature, OF [0C] 50 to 75 [10 to 24] Limit pressure differential across caprock (1)
Temperature cycle time, months 3 to 6 Eliminate external casing packers (1)
Maximum BHP, psig [MPa] 1,000 to 2,000 [6.9 to 13.8] Torque-turn monitoring of connection makeup (1)
Minimum BHP, psig [MPa] <500 [<3.4] Recondition well (1)
Pressure cycling time, months 3 to 6 Annulus rupture disks to relieve pressure (1)
Well leakage 0 Casing patches over collar leaks (1)
Probability of leakage, % 7
Leakage rate if observed, ft3/D [m 3/d] <5,000 [<141.6]
Time before leakage observed, weeks 0 to 2 'Numbers in parentheses represent the number of times this response was given.
Cycles before leakage observed 0 to 3

JPT • November 1989 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 1 1 5 1

 TABLE 4-EFFECT OF CURING PRESSURES
ON BOND FAILURE (from Bearden et al. 3 )
.......k -
.......-
Curing Pressure Bond Failure •
~
Annulus Internal Pressure I /0/
I
/;
Test psi MPa psi MPa ~ MPa

0 800 5.5
V
0 0 0

2 0 0 4,500 31.0 200 1.4

. .000
/ i
I
I°

3 1,500 10.3 0 0 1,500 10.3

1000
0
- BOND
i
0200

FAILURE
.
PRESSURE. psi.
.... •
&GO

4 4,500 31.0 0 0 3,100 21.4

Fig. 25-Effect of curing pressure on bond failure (from Bearden
et al. 3 ).

may provide greater statistical differentia- The survey results do not confirm this as- will help increase bond strengths. Second,
tion of the data. sumption (Figs. 23 and 24). In 421 wells some leakage can be expected at the higher
The time period in which the wells were with known leakage and a known number pressures, and well completion planning, in-
drilled was studied first by comparing the of cycles until leakage was detected, 90% cluding the possible use of mechanical seal-
percentage of wells reported as having leaked in zero to three injection/withdraw- ing elements in the annulus area, is critical
leaked during the five different time peri- al cycles. In addition, for 426 wells having for successful sealing.
ods. No statistical difference (defined at the known leakage and a known period of time
95 % confidence level) could be found be- until leakage was noted, 77% leaked in less Cementing Techniques. The data show that
tween time periods. than 30 days (most typically in less than 2 the cementing methods, techniques, and ma-
Given a lack of evidence that time period weeks). Leakages were reported at greater terials used in the wells surveyed vary wide-
is a highly significant factor, additional anal- than 10 cycles in only 2.6% of the cases. ly and that many of the recommendations
yses were conducted with data from all time In low-cycle fatigue of most engineering from the past 20 years of cementing research
periods combined. For this reason also, materials, failure is the result of operating were not used (most of the wells surveyed
Figs. 1 through 24 are presented without at a stress level that is too near the materi- were drilled before 1965). Many of the leak-
differentiation by period. al's ultimate limit. These may be thought of age problems reported can probably be at-
Of the variables considered, only depth as quasistatic. tributed to this. This statistic also implies
and maximum pressure showed statistical In our case, failure may be the result of that the leakage integrity of new wells can
significance. Wells that leaked tended to be operating too near the static bond strength be improved immediately by use of up-to-
deeper (in excess of 4,500 ft [1372 m)) and of the system. Bearden et al. 3 showed that date techniques and additional quality-
have higher pressures (>3,000 psig [20.7 static bond strengths are generally equal to control measures.
MPa)) than those that did not. or less than the pressure applied during cur- Fig. 10 shows that in 79% of the cases
Finally, subjective responses were ing (Table 4 and Fig. 25). Because mud reported (3,297 wells), cement setting times
cataloged according to what were thought weights typically are chosen to apply an an- were 25 hours or less. Because the struc-
to be the most important factors causing nular pressure that is just greater than the tural properties of cements are known to in-
leakage and how to prevent or stop it. These BHP to be contained, static-bond-strength crease significantly over time, longer curing
are shown in Table 3. The causes mentioned values may very often be close to BHP's. times would be beneficial. This is especial-
most often for annular leakage were Early failures by either static or low-cycle ly true for the poz-mix cements used in 36%
microchanneling of the cement, generally fatigue loadings would then be expected. of the cases reported.
poor cementing jobs (most often on wells Bearden et al.' s work also shows that Bearden et al. ' s work showed that the ce-
drilled before 1965), and failed wellhead there is a general leveling off of bond ment should be cured with the highest pos-
seals. The most frequently used remedial strength with increasing annulus pressure, sible hydrostatic pressure applied across the
methods for controlling leakage were to per- so it becomes harder to generate adequate cement annulus only (highest possible pres-
forate the casing and squeeze cement at the static bond strengths as BHP increases. The sure differential across casing wall). Our
location of annular leakage, to run additional frequency ofJeakage failures would be ex- survey did not ask for this information, but
liners, and to repack faulty wellhead seals. pected to increase in these cases. one major cementing contractor and sever-
Discussion of Results Both hypotheses are supported by our sur- al operators of both production and gas
vey data, which show that the only statisti- storage wells were contacted and reported
Frequency of Failure. The data show that
cally significant variables correlating with that this is not a common practice.
for most of the gas storage wells currently
leakage are depth and BHP. Wells with As mud is displaced through the casing
in operation, surface leakage for whatever
BHP's of 3,000 psig [20.7 MPa] and great- into the outer annulus, the displacing pres-
reason is not a significant problem. Of the
er are near the level where bond strengths sure is typically held for some period of time
wells reported in operator surveys, 95 %
rapidly start to level off. (zero differential) before being released. Af-
have no known surface leakage; of those that
do, most report relatively minor leakage
n is also significant that cyclic fluctuations ter release, there is a differential across the
were mentioned by only one respondent as casing, but it varies widely because of the
rates.
a cause for leakage, even though all respon- cement weight and column height. A hydro-
Mode of Failure. The basic assumption un- dents were told that this was the subject of static head is occasionally, but not common-
derlying the project was that periodic tem- the research program. It was listed among ly, applied across the annulus (in addition
perature/pressure fluctuations of gas stor;ige the minor causes for leakage. to that provided by the mud itself).
wells subject annular components to a 'fa- This has several major implications. First, And finally, the data cast doubt over the
tigue situation, causing leakage later in the increasing the annular pressure differential original assumption of a fatigue action that
life of the well that is not evident early in during curing, through either higher mud causes significant leakage problems. While
its life. weights or supplementary annular pressure, the presence of such a mechanism remains

1152 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - November 1989 • JPT

reasonable from a common-sense approach, 5. Carter, G. and Slagle, K.: "A Study of Com-
it is being masked, if present, by more fun- pletion Practices to Minimize Gas Commu-
damental considerations. It is these fun- nication," IPT (Sept. 1972) 1170-74.
damental considerations that should be given 6. Carter, G.L., Cook, C., and Snelson, L.:
"Cementing Research in Directional Gas
first attention.
Well Completions," paper SPE 4313 present-
ed at the 1973 SPE European Meeting, Lon-
Cement-Column Height. Although the data don, April 2-3.
are not conclusive, the results from two 7. Stone, W.H. and Christian, W.W.: "The In-
programs 2 ,26 suggest that leakage may ability of Upset Cement to Control Forma-
occur more readily in shorter cement tion Pressure," paper SPE 4783 presented at
columns. In iliese tests, leakage occurred at the 1974 SPE Formation Damage Symposi- Marlow
significantly lower pressures than in other um, New Orleans, Jan. 31-Feb. 1.
tests performed on longer models when the 8. Christian, W. W., Chatterji, J., and Ostroot, Roy S. Marlow, principal engineer at
models were similarly prepared. This im- G. W.: "Gas Leakage in Primary Southwest Research Inst.'s Dept. of
Cementing-A Field Study and Laboratory Structural & Mechanical Systems, works
plies that increasing the length of the cement primarily in development of drilling and
Investigation," IPT (Nov. 1976) 1361-69.
column may increase the statistical chances 9. Garcia, J.A. and Clark, C.R.: "An Investi- production tools and tubular products. He
of obtaining a good gas-blocking bond at gation of Annular Gas Flow Following Ce- previously worked for Reed Tool Co. and
some point. This theory contradicts one the- menting," paper SPE 5701 presented at the for the Texas A&M U. Research Founda-
ory that says cement columns should be kept 1976 SPE Formation Damage Symposium, tion. He holds an MS degree in mechani-
as short as possible to minimize the amount Houston, Jan. 29-30. cal engineering and an MBA degree In
of slurry that can revert to mix-water gra- 10. Cook, C. and Cunningham, W.C.: "Filtrate management.
dient (and the resultant loss of hydrostatic Control-A Key in Successful Cementing
head). 27 Although Sutton et ai. 1 pointed Practices," IPT(Aug. 1977) 951-56; Trans.,
out that this has not been proved, the use AIME,263. 23. Cooke, C.E. Jr., Kluck, M.P., and Medrano,
11. Levine, D.C. et al.: "How to Prevent An- R.: "Annular Pressure and Temperature
of shorter columns (such as for stage ce-
nular Gas Flow Following Cementing Oper- Measurements Diagnose Cementing Opera-
menting) remains popular throughout the in- ations," World Oil (Oct. 1980) 85-94. tions," IPT (Dec. 1984) 2181-86.
dustry. The answer may lie in a trade-off 12. Cooke, C.E. Jr., Kluck, M.P., and Medrano, 24. Sutton, D.L., Sabins, F., and Paul, R.: "New
between these two requirements. It may be R.: "Field Measurements of Annular Pres- Evaluation for Annular Gas Flow Potential,"
true that shorter intervals are better up to sure and Temperature During (and After) Oil & Gas 1. (Dec. 17, 1984) 109-12.
a certain point that is not approached in ac- Primary Cementing," IPT (Aug. 1983) 25. Marlow, R.S.: "Effects of Temperature/Pres-
tual field completions. 1429-38. sure Cycling on Cement Bonding Character-
Researchers have also shown difficulty in 13. Griffin, T.J., Spangle, L.B., and Nelson, istics in Gas Injection/Withdrawal Wells,"
obtaining adequate bond strengths in labo- E.B.: "New Expanding Cement Promotes final report, AGA Project No. PR-15-529,
Better Bonding," Oil & Gas 1. (June 25, American Gas Assn., Arlington, VA (March
ratory tests. While this difficulty probably
1979) 143-51. 1987).
results largely from differences in labora- 14. Tinsley, J.M. et al.: "Study of Factors Caus- 26. Pittsburgh Testing Laboratory, technical
tory procedures, the procedures are repre- ing Annular Gas Flow Following Primaty Ce- report No. 5158 (March 3, 1967).
sentative of those used in the field and agree menting," IPT (Aug. 1980) 1427-37. 27. Levine, D.C., Thomas, E.W., and Bezner,
with field experience. 15. Watters, L.T. and Sabins, F.L.: "Field Eval- H.P.: "Annular Gas Flow After Cementing:
If bonding is not as good as suggested by uation of Method to Control Gas Flow Fol- A Look at Practical Solutions," paper SPE
our survey, a significant amount of leakage lowing Cementing," paper SPE 9287 8255 presented at the 1979 SPE Annual Tech-
may occur between formations that is not presented at the 1980 SPE Annual Technical nical Conference and Exhibition, Las Vegas,
being accounted for at the surface. Conference and Exhibition, Dallas, Sept. Sept. 23-26.
21-24.
Acknowledgments 16. Webster, W.W. and Eikers, J.V.: "Flow Af- 51 Metric Conversion Factors
ter Cementing-A Field and Laboratory
The work described was funded by the Study, " paper SPE 8958 presented at the ft x 3.048" E-OI = m
Storage Reservoir Supervisory Committee ft3 x 2.831 685 E-02 = m3
1980 SPE Unconventional Gas Recovery OF (OF-32)/1.8 = °C
of the Pipeline Research Committee, Ameri- Symposium, Pittsburgh, May 18-21. gal x 3.785412 E-03 = m 3
can Gas Assn., and represents a condensa- 17. Sabins, F.L., Tinsley, J.M., and Sutton, in. x 2.54" E+OO = cm
tion of the final report for Project No. D. L.: "Transition Time of Cement Slurries Ibm x 4.535 924 E-Ol = kg
PR-529, March 1987. I thank the Ameri- Between the Fluid and Set States," SPEJ psi x 6.894 757 E+OO = kPa
can Gas Assn. for funding the project and (Dec. 1982) 875-82.
'Conversion factor is exact.
18. Sabins, F.L., Sutton, D.L., and Cook, C. Jr.:
for permission to publish the results. "The Effect of Excessive Retardation on the Provenance
Physical Properties of Cement Slurries," IPT
References (Aug. 1984) 1357-65. Original SPE manuscript, Effects of Tem-
1. Sutton, D.L. and Sabins, F.L.: "Annular Gas 19. Sabins, F.L. and Sutton, D.L.: "The Rela- perature/Pressure Cycling on Cement
Flow Theory and Prevention Methods De- tionship of Thickening Time, Gel Strength, Bonding Characteristics in Gas-Injection/
scribed," Oil & Gas 1. (Dec. 10, 1984) and Compressive Strengths of Oilfield Ce- Withdrawal Wells, received for review
84-92. ments," SPEPE (March 1986) 143-52; July 15, 1987. Paper (SPE 17121) accept-
2. Carter, L.G. and Evans, G.W.: "A Study of Trans., AIME, 283. ed for publication Dec. 21, 1987. Revised
Cement-Pipe Bonding," IPT (Feb. 1964) 20. Bowman, G.: "Cement Liners Successfully manuscript received June 15, 1988.
157-60. Through Gas Thief Zones, " World Oil (Jan.
3. Bearden, W.G., Spurlock, LW., and 1983) 113-19. JPT
Howard, G.C.: "Control and Prevention of 21. Cheung, P.R. and Beirute, R.M.: "Gas Flow
Inter-Zonal Flow," IPT(May 1965) 579-84; in Cements," IPT (June 1985) 1041-48.
Trans., AIME, 234. 22. Cooke, C.E. Jr., Kluck, M.P., and Medrano,
4. Scott, J.B. and Brace, R.L.: "How Coated R.: "Field Measurements of Annular Pres-
Casing Reduces Primary Cementing sure and Temperature During Primary Ce-
Failures," World Oil (Aug. 1966) 66-69. menting," IPT (Aug. 1983) 1429-38.

JPT • November 1989 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - lI53