society of Petr*

*ers

I

SPE 30012 Completions and Stimulations for Coalbed Methane Wells Ian Palmer.* Amoco. Hans Vaziri, Technical University of Nova Scotia, Mohamad Khodaverdian,’ TerraTek, J;hn McLennan,* TerraTek, K. V. K. Prasad,’ Amocoj Paul Edwards,* Amoco, Courtney Brackin, Amoco, Mike Kutas, Amoco, Rhon Fincher, Amoco

* SPE Members

Copydgfit 1SS5, Society of Petrdaum

Engineers, Inc.

Thii paper wee prepared for presentation at the Intemetional #kafing

on Patmfaum Engineering held in Ssdjing, PR China, 14-17 Nwambar

1S95

TM paper wee aa4actad for pmaentation by an SPE Program Cwnmittaafolfcwhg review of informatii mntahad in an amtract submitted by the author(a). Contents of the paper, se pfSWnt8d, have rmt bean reviewed by the Socisty of Patmfaum Engineers and am aub@tad to mrmctmn by the author(s). The material, as pmaantad doaa not nacaaaarity mflaot any poaMn of tha %ciaty of Petmfaum Enginaara, its officers, or mambara. Papers prwentad at SPE maatiis am aubjact to publiitbn review by Editorial Oommittaaa of tha Socialy of Petroleum Engineers. Penmiaaion to COPYis restricted to an atmtracf of not more than 300 words. Illuatrafiia may not be copied. The abstract should oonfain conapicuoua aoknowkrdgmant of wtwm and by whom the paper is pmaantad Wdta Libradan, SPE, P.O. Sox S33S3.S, Richardson. TX 750%3--, U.S.A. [Fs6*ifiik 2$~S=-W%

Abstract Amoco is producing coalbed methane from several hundred wells in both San Juan and Warrior basins. These wells were completed/stimulated in one of two ways: (1) openhole oavity completions, (2) hydraufic fracture stimulations through perforations in casing. Cavity operations are described, and new data from several cavity completions is presented and analyzed. The latest geomechanics modeling of the formation of cavities in coalbeds is presented. The model allows the growth of a cavity as tensile failure occurs, and computes increases in permeability in a stress-relief zone that extends tens of feet from the weii. Critical parameters are given for the success of cavity completions. A pulse interference analysis is discussed: as wefl as intefwell permeability, this can provide information on stressdependent permeability. Finally, some wells which were originally cavitated did not perform up to expectation, and have been reoavitated with remarkable success - these are also examined. Amooo has tried several different kinds of hydraulic fracturing treatments. Results of comparisons between foam fracture, slick water fraoture, and gel fracture treatments are presented. a.-.: al--l -- -.--I -..-.. . . .&.-- *rw r-;~-~ -e thida mf *- f ir. wm~al IWI I= 01 e WJ=I I IUl r ~lwl 1= UUW,VV”, .,A, =MLISWGUI way zone in the San Juan Basin. In the Warrior Basin, water fracture treatments with and without sand have been compared. Lastly, foamed water cleanouts, without sand, have been deployed, and their success is reviewed.

Introduction h-f this paper we present new information on completionaktfmulations of ooalbed methane wells. Specifically, we discuss (1) openhole cavity completions in the fairway (sweet spot) of the San Juan Basin (Colorado and New Mexioo - see Figure 1), and (2) fraoture stimulations in the San Juan Baain and the Warrior Basin (Alabama).

Cavity Operations The openhole cavity completion has been used with tremendous suooess in the San Juan Basin. 1.7,13,14,17some wells produce in excess of 10 MMCFD from only 3,000 ft depth in the

fairway zone (Figure 1). In the cavity operation, a series of injections (or shut-ins) and blowdowns (aotually, a controlled blowout) is performed over typically a two-week period. Coal fails and sloughs into the wellbore, and is ejeoted from the well, leading to creation of a cavity (enlarged wellbore). A plastic or shear failure zone is also formed beyond the cavity, and in this region the permeability is changed. A typical Amoco oavity operation was described previously.”7 Below is an elaboration of certain aspeots of cavity operations in the San Juan Basin fairway 1. The openhole potion oi the weii is generaiiy 200-300 ft hi height, containing usually more than 50 ft of net coal. The coals are divided into the basal coals, which are usually the more productive, and the upper coals. Normally 7-in. casing is topset above the top coal, and TD is only a couple feet below the bottom coal. 2. A typioal cavity operation entails a sequence of(1) cleanout of the well in the evening using air (1,500-2,200 SCFM) and water (20-100 BPH) inj~lons, followed by (2) flow testing lasting typically four houra, followed by (3) cavity operations fCii

~~~),

t’@~ibj

+!

O ~~~fie.

u,-

AII+m W-B, .- th= ...” rf.utim~ ““,.....”.

RafQra -=. .-

the flow test and CST, the bit ISeither pulled into the casing shoe or to the surface. The sequence is repeated many times over typically 10-20 days. 3. All flow teats are conducted through a 3/4 in. choke, typically for four hours. All pressure surgings are conducted by rapidly opening a surfaoe valve, allowing gas and water and coal fines to be expelled through blooie lines to the pit. 4. The basal seams seem to respond more than the upper seams to the cavity operations, presumably because they are more friable. 5. It is not uncommon to see 0.5-1 in. pieces of coal come to the surface during cavity operations. 6. In flow tests in good wells, flows during the cavity operations often decrease wirn time over i-4 iirs. This may be the transient effect that is predioted by the oavity modelin~ (see later in this paper). This contrasts with flow tests in the

.-

. COMPLETIONS AND STIMULATIONS FOR COALBED METHANE WELLS

2

I law

I

R16W

I

I ntaw I

n7w

~~~”

I

I Rsw I

I Rsw I

I

“

I

-~

I Rlw

“1

T24N

T2SN

#

-1

.-’

NEW

M2xlqo -1

T22N

.--L<-.

$ii JUAN

..-

co

T=

\

) -f

1

\ I ● M9

..-

I

I

nlw

I

—

..9?-

pduoson

m

Raw

mm

(Mew)

*b,m

*WOO

5 :.:.~.: .........

o

.:=

0

s

Figure 1. High productivity

12 ml

-

lokm

culiow

InlwvAl Vuhek

-7s6s

fairway (awaat spot) in the San Juan Basin (after ref. 18).

tighter formation south of the fainvay, where ratea generally increase with time. 7. Pressure surges are of two kinds: natural and injection. Natural surges are those in which the well is shut-in for about an hour before biowaown. injaotion surges are tinose in which air (typically 3,000 SCFM), sometimes combined with 20-50 bbl water sweeps (typically 2-3 BPM), is injected during a period of about an hour before blowdown. 8. Wells are cleaned out to TD, then a flow testis conducted, followed by pressure surges (CST). When the bit is tripped back into the hole, an estimate of the fill in the well is normally made. The bit may tag fill all the way to the bottom, or a bridge of fill of limited height, or even a sticking ledge. Thus, the fill records are probably sometimes upper limits to the aotual net fill of coal fines in the wellbore.

shown above the line by the open bars (air only), and by the hatched gray bars (air + water) in the plots. The number associated with each bar is the number of surges conducted during that sequence. When there are gaps of a day or more between surgings, these are normally prolongad cleanout times, due to difficult drilling conditions (running coal oausing stioky drilling, and sometimes even stuck drillpipe; very often heavy coal returns are obswed at the surface). These times appear to correspond to continuing coal failure, when the coal is very friable, while a full head of water is exerting pressure at the bottom of the hole (this can happen if the cohesion is low enough2).

CAVITAM

9. Sometimes when the hole is unstable, the well files indicate the bit appears to be drilling new hole. 10, During cleanout, greater coal returns are achieved using more viscous fluids: (1) water sweeps, (2) soap sweeps, (3) polymer sweeps (the latter are rarely used by Amoco).

11. Shale particles are often expelled along with ooal fines during cavi operations. For example, on= well reported 70% coal fines and 3W0 shale; another well reported 5W0 coal fines and 50% shale. Clearly shales fail along with the coal, and the “bmk case” model previously espoused at the COAL site,’ may not be defensible.

so 43 tooo 4.0

il-

3.3 3.0

i2000

2.3

1!

2.0

0

M 1.0

-2000

0.3

Data from Cavity Completions

-4000

0.0 1234

Nine other Amooo oavity completions and reoompletions have been summarized,l 1 and some of these are shown as Figures 24 These are all from the faitway in the San Juan Basin, at typioel depths of 3,000 ft. Natural surging is shown by the black bars below the line on the plots. Injeotion surging is

584

S67S

91011121314

1516171819

~2122z12423

O’W

Figure 2s. Surging and gaa flow rata during original cavity oomplatfon.

1.PALMER, H. VAZIRI, M. KHODAVERDIAN, J. MCLENNAN, K. PRASAD, P. EDWARDS, C. BRACKIN, M. KUTAS, R. FINCHER

‘SPE 30012

some values are upper limits (shown by an arrow pointing downwards). Finally, the maximum surface pressure after typically one hour of natural buildup, or air injection buildup, is also plotted.

CAV17A71CN

m

.90

1620

-20

1600

-m

1400

We make the following observations:

-60

1. Character of flow rate increase There is a wide variety here, from the classic case of a rapid increase which then flattens, as in Figure 2, to cases in which the rate accelerates at late times. The former case is reminiscent of the COAL site reported previously, 13,14,17 and gucCegSfUll)/ modeled below. The latter case is exemplified by Figure 3, where the rate does not change much over the first five days, but then increases to 3:1 over the ensuing six days. Furthermore, in one area, cavity operations were carried out for around seven days before the coal failed significantly. In the end, the wells produced more than 1 MMCFD (compared with wells with cross-linked fractures that produced very little). Perhaps virgin cohesion was higher in this area, and persistent surging eventually lowered the cohesion enough to fail the coal.

1200

--::

Kilo m

-30

600 a

-m

200

-10

0

*O 1234567891011121314

1516171819

3

Z12122232423

-

Figure 2b. Maximum of pressure buildup before blowdown, and fill in wellbore (corresponding to Figure 2a.) CAVITAM

In another 0sss, the flow rate initially increased by about a factor of 3 in the first three days, then increased by another factor of 1.2 over the next eight days, before increasing by only another factor of 2 over the next eight days. These final eight days are again interesting because there was no pressure surging, although coal returns continued - apparently the well cavitated during circulation, implying very friable coal of low cohesion. Note that cohesion is defined as residual shear strength, when normal stresses are removed.

4.0 2.6

In one other case, shown in Figure 4, the flow rate increases fairly steadily with time over about 22 days. During this time, natural cavitations occur every few days, interspersed with occasional natural surges. The measured fill decreases fairly steadily over the course of the cavity operations, being initiefly high (120-200 ft of fill), decreasing !Q Q410ft Qffi!! n~ar ?he end of ?h’eqeratiml:

0.6 0.0 1 2

34S

67

S910111213141S

161718192021

22 Z3Z23

m

Figure 3s. Surging and gas flow rate during original --..la..

----1-*1--

Guvny

CxJlrlpluuvll.

MCAVrrA~

2oa1

I

I

1

6000

3.0

6000

23

I

2.0 to I

~2002

M # o

1.0

-moo

0.3

II

4

-aoo

0.0 23

1 23

4567891011121314151617

I

4S

678

910111213141S

16171819

~21==

242S26

181920212223M23 m

Figure 4a. Surglngs and gaa flow rata during a cavity racavitstion.

Figure 3b. Maximum of pressure buildup before blowdnyn. --. .. and ---- fill .... In .. . wallbena . . . ... . . . kemesea ,--. .--r -..nd!n~

t~

Figure 3s.) Plotted afeo are the gas rate increases, relative to the first flow rate measurement during the cavity operation. These rates are obtahd by four-hour flow tests through a 3/4 in. choke. Note that if the gas is wet, the gas flow rate will be an overestimate (shown by a minus sign in the circle plotted). Also plotted is the fill observed after each surging operation. As explained above,

585

2. Magnitude of flow rate increase: Flow rate increases of 4:1 are common. Only 3/9 oases have flow rate increases lees than 4:1. The greatest flow rate increase was an original cavitation, which increased by 81, and over this 48-day period there were not many surgings, but long periods of cleanout, presumably accompanied by continual coal failure.

. COMPLETIONS AND STIMULATIONS FOR COALBED METHANE WELLS The corresponding shear failure zone has a radius of 34 ft. Cohesion is the inherent shear strength of the coal. A value of 15 psi is much lower than has been adopted in the~ast. This value has been substantiated by laboratory testing. Note that .- *Ltine faiiure surface in me latest muurmrlg IS UIIII IUiII, ~fi~ tk 15 psi mhesion is an extrapolation of the high stress linear portion (the low stress linear portion has a cohesion of zero).

RECAVITATION

,-----

1234567891011121314

1516171819

A-I:--

:-

b.:!:

----

. There is a zone of permeability enhancement immediately beyond the cavity. At the COAL site an enhanced permetillity in excess of 500 md is found at the cavity surface (virgin perm = 25 red). Further out, however, is a zone where perme ability is reduced below virgin levels. An effecthre skin factor of about S = -3.5 was computed at the end of the first day of cavity operations at the COAL site (from the enlarged wellbore, and the enhancedhduced permeability zones).lg This can be compared with that calculated from a perme~llity profile obtained by ARI history matching S = -5.9.10

Z121Z2Z32423

m

Figure 4b. Maximum of pressure buildup before blow down, snd fill in wellbore (corresponding to Figure 4s).

●

An upper limit to the titmulation was considered by removing the zone of reduced permeability, and replacing it by virgin permeaMity.3 The results area more neq~tive skin (S = -4.7), ,- L-A -- --------~ ..4A *L.- A Dl .-a, ,146 m oener

3. Maximum recovery following blowdowns: Maximum pressure buildups have been recorded over roughly one hour during (a) natural surging and (b) air injection surging. In the case of Figure 4b, it can be seen that air injection pressures exceed natural shut-in pressures by 300400 psi. This is as expected. Note also that the maximum pressure buildup during the natural shut-ins falls slowly over the 25 days of cavity operation, perhaps dropping by 100 psi. This is consistent with an increasing zone of enhanced permeability. Finally, we note that in another case, the diacrep ancy between injection and natural pressure buildup is quite a bit larger (700-800 psi) than in the case of Figure 4b (300-400 psi). This must reflect the permeability of the formation, and suggests that the perm is larger in the case of Figure 4. Curiously, gas producing rates after the respective cavity operations were about the asme, and therefore do not support this interpretation. 4. Based on nine cases, bottomhole measure buildups after one hour in natural surgings average .28 psi/ft. “For air injection surgings, the average is .45 pai/ft. In all cases, air injection pressures are less than .59 psi/ft, which is lower than any virgin horizontal stresses measured in basal seams. This should imply injection below parting pressure, but the parting pressure itself may be lower due to stress relief around the wellbore. For air and water injections, bottomhole injection pressures are .80-1 .14 pai/ft, assuming a full hydrostatic column of water, and these all exceed the average minimum stress in the basal seams, implying fracture injection. The water injection appears to assist in initiating coal failure.

Historically, cavitation operations have relied on empirical field obsewations and local @@erience.This has evidentiy been a successful, although not optimized, approach in the fairway. To be able to optimize cavity operations, to provide a basis for recavitation decisions and to apply the technique to a greater geographical area, numerical modeling has been undertaken. The following points are a summary of published worl@3 and recent unpublished work. The simulations performed to date have successfully matched the size of cavities, as measured in the field. Example at the COAL site, in the northwestern end of the fairway, a 94t diameter cavity has bean matched by using a cohesion c’ of 15 psi.

586

dgrwmurn

WILI I LIIW

mnt ItxuIu.

✎

From beginning to end of Day 1 of cavitations at the COAL site, the gas production improvement wee by about a factor of 3 in the model. However, in the field, the production improvement was about a factor of 5. This may also suggest the zone of enhanced permeability in the field is larger than that computed by the model. Alternatively, the stress-permeability function could be stronger than the one adopted.

✎

The larger the cavity and the plastic failure zone, the greater the extent of enhanced permeability, the smaller the effect of the zone of reduced permeability, and the greater the gas production.

✎

Continuation of surging: A second day of surging was added to the previous modeling of the first day. The second day had only a minor influence on failure cavity radius increased from 4.5 to 5 ft and shear failure radius increased from 34 to 40 ft. Clearly, the first day was of paramount importance, at least at the COAL site, and the gas production rates reflect this.

●

Influence of wellbore cleanout a 24-hr period of open flow before the second day of surging increased the effective stress (due to pore pressure drop), and made it harder to fail the coal during the subsequent surges. This suggests cavitation would be easier and more efficient if the cleanout wriod could be reduced.

✎

An overpresaured formation may not be required, as the modeling indicates that a cavity will form in a normally pressured formation, if the cohesion is low enough.

✎

Cavities can be created merely by circulating the well underbalanced, if the cohesion is low enough. This explains the “natural” cavity technique used by Meridian (i.e., no slowdowns). However, blowdowns do appear to exacerbate the failure of the coal.

Summary of Reoent Modeling Resuits

●

—-

✎

✎

Young’s modulus does not have any influence on the size of the cavity in the coal (at least for values between 30,000 end 450,000 psi). The geometry of the seams (thickness and depth) is not a criiicai factor in determining the’extent of cavitation. “However, as depth increases, so does in-situ stress, and the shear failure xme,“ rlna~ lamer -“, “--” htaenme -------.-. =-. -

Laboratory Resuits in Support of the Modeiing Field obsenmtions and numerical simulations were supple mented by laboratory measurements, particularly to aeeese w-

s

SPE 300.I 2

5

1.PALMER, H. VAZIRI, M. KHODAVERDIAN, J. MCLENNAN, K. PRASAD, P. EDWARDS, C. BRACKIN, M. KUTAS, R. FINCHER

meability and cohesion values. Some of the relevant obsewations areas follows: . Laboratory experiments have measured cohesion in a coal core under triaxial stress, but these are substantially higher (few 100 psi) than the cohesion implied by the model (c’= 15 psi at the COAL site). However, careful experimentation at low confining stress levels substantiate a bilinear or parabolic failure envelope consistent with iower vaiuee of cohesion. in addition, direct shear testing on coai samples has specfficsily supported iower vaiuee for cieat+ontrolled cohesion (c’ = 30 psi), in cioeer agreements with tnose impiied by the numerical modeling.3 ●

Laboratory work under triaxiai stress has shown that permeability increases when coai faiis, by about a factor of 3. Furrner, recent direct shar mw~urerlleil’~ ifi~~te ~~~~ permeatitity increases much more and these results have been incorporated into the modei~ in fact, the permeability may increase even beyond the shear faiiure zone, due to shear displacement and diiatancy.

✎

A large coai biock has been succe&sfuiiy cavitated, and the resuk wiil be reported eisewhere.l 1 A startiing observation from this tasting invoived the infiuence of water injection. After a pseudo-stable cavity was created by cyciic gas injtilon and biowdown, injection of a water stage reinitiated unstabie weiibore conditions and additional hoie enlargement.

Criticai Parameters for Cavitation to Succeed Our preliminary assessment is that cohesion is the most crftical parameter in determining cavitation. Another impotint parameter is the pore pressure gradient near the surface of the cavity. This is a consequence of (a) iargelrapid pressure changes at the wellhead, and (b) formation permeability. A third parameter of some impcxlance is the in-situ stress condition, as mentioned above. Uniese noted otherwise, the base case reeuite presented below assume a cohesion, c’, of 15 psi, initial permetiiiity, k, of 25 md, and permeability multiplier of 10 in the piaetic (shear faiiure) zone to account for diiation-induced improvement in permeability. The surging sequence adopted for ali the cases is shown in Figure 5. ,, .,.., X6 i

2 -1100

Implications 1. One of the most significant contributions of the numericai modeling has been in back-calculating parameters and properties that are difficuit to determine by direct meaeuremente or from laboratory experiments. For instance, the numericai modeiing has furnished values for iarga-ecaie cohesion. 2. Cavity etabiiization: Modeiing reeuits suggest that most of the coai faiiure occurs in the first two days of cavitation, at ieast at the COAL site. However, in many fieid applications, coai fines are often recovered during many days of cavitation (up to a month). This may reflect (1) iower cohesion coal that requires a iarger cavity before it statiiizee, (2) the heterogeneity of the coai, and perhaps “patches” of iow COhAOn ma! that are continually encountered as the cavity grows in size. The stabiiity qu~lon has to be determined in the fieid, by how iong it takes to reduce the fines recovered to a iow ievei. This, of course, iowers the chance of fines continuing to come after the weii is on produtiion, and plugging up the formation or causing ffii in the weiibore. 3. The cavity compi~ton technique appears to have a typical skin factor of -3.5 (or -4.7 upper iimit) which does not indicate a tremendous stimulation effect. Nevertheless, in the fairway openhoie cavities connect better to the reservoir than do hydrauiic fractures. This is probabiy because fractures cause subetantiai damage to the reservoir, e.g., damage due to driliing mud, cement invasion, geiied fracture fluids, fracture stress, and coai fines (and the iaet may weii be the most important).l 4. A “stress threehoid” effect has been espoused, 12 whereby in the fairway a minimum stress is needed for cavities to be succesefui. This is based on fieid observations that are interpreted to show stresses in the fairway are 100-150 psi greater than outside the fairway. Howaver, in our modeiing we do not see sensitivity to stress anywhere near as strong as this.

587

-1500

0

1

5

4

3

1

Tim od

Figure 5. Applied wefl pressure for the COAL she problem (s110ssss). The size of the cavity and plastic zone, and the gas production, are strongiy controiiad by the cohesion of the formation. At the COAL site, ac’ of 35 psi yieids oniy about a fourth of the produtilon of the beet-fit c’ of 15 psi (see Figure 6). As cohesion becomes smaiier, the cavity and piastic faiiure zone, as weil as flow rate, become iarger, as aieo shown in Figure 6.

m o 0

5

10

15

20

25

m

35

@

Cdlsial (pi)

Figure 6. Dependency of failure and flowrate on cohesion (tensile = cavity radius; pfsstfc = sheer failure zone radius). The voiume of coai washout at the end of driiiing (not cavitation) can provide a good indication of the cavitation potentiai of the formation. This is iiiuetrated in Figure 7 where for the case of c’ = 2 psi, the cavity radius is about ~.55 ft. whereas for the case

.

. COMPLETIONS AND STIMULATIONS FOR COALBED METHANE WELLS

8

of c’ =35 psi, the cavity radius is about 0.75 ft (note that the initial cavity radius is 0.5 ft). A caliper can be used after drilling, to measure the washout diameter. 1.6 1. 1 1 I

35 psi, there is in fact a siight reduction in the flow rate at the end of surging that is caused by development of higher stresses outside of the piastic zone that act to reduce the overall permeaMl_c=15pi

~c=2pi

t..l

0.6

o

. . . . . . . . . . ..l 5

10

20

M

23

302340

c (pi)

Figure 7. Variation of cavity enlargement with cohesion at the end of drtliing. Figure 8 shows the variation of flow rate with surging as a funtilon of cohesion. Negative flow rates mrrespond to flow from the well. Positive rates correspond to injtilons into the well. Note the transient nature of the flow rates predicted by the model: a sharp peak lasting <10 min followed by a decline over 1-2 hours (see decline after last surge). This transient behavior is supported by field data presented earlier for wells of higher permeability. Note that for the case invoiving c’ = 2 psi, the flow rate improves with every cycle of pressure surge, whereas for the case of c’ = 35 psi, there is no net improvement in production rates. —c. 20

~,.

_c=35psi

..,,,

=2p9i ,,.

-

— c=15pai

, . . . . . . ..

A...

. .

0

m

10

60

50

40

30 mid -

(iv

Figure 9. Comparison of permssbiiity at the end of surging sea funotion of ooheeion. 6L . . . . . . . . . . . . . . . . . . . . . . . . . . . ..r.

... ...4

~,,,a,

t\

c=35p5i

. . . . . . . .. . . . . . . . . 4

-0

/

5

10

15

20

25

303540

Cohesion (psi) Figure 10. Refathre flow rate increase from drilling to end of surging, as a function of ooheeion. Sensitivity to virgin ei+a --”. a~al!mad “r.” 1.-

#it

’11

L

o

1

2

3

4

5

6

I

4

7

8

T3me(IWI)

Figure 8. Fiow rate as a function of cohesion. The perm enhancement is a strong function of cohesion. Figure 9 illustrates the radial perm profile at the end of surging as a function of cohesion. Figure 10 shows the relative increase in flow rate between the end of drilling stage and the end of surging as a function of cohesion. Aiiough the absoiute fiow rates are not correct, because there is only singbphase fiow in the model, this relative flow rate increase appears to provide a more valid comparison with field data. The most significant observation to make is that stress relief resulting from the wellbore mmpltilon can dramatically improve the production rate if cohesion is iow. For the case of c’ = 2 psi, the increase is more than five times, and for c’ = 15 psi, the increase is over three times. For c’ =

permeabilitythe match to the

a uimin narmanhilitu nf ~~ rn~m Whan Q

---GOAL

thi~ k

replaced by 2.5 m~l’t%’e&~~~~u~&aeed fo~ 4~to6~ ~ and the shear failure zone increased from 34 to 48 ft radius (see Figure 11). The reason is that a iower virgin permeability results in a steeper pressure gradient near the cavity surface, which in turn exacerbates the failure. Figure 12 shows the variations in flow rate during surging for all the permeability cases. Of course, the higher the initial permeability, the iarger will be the production. However, what is interesting is the relative increase in production. As depicted in Figure 13, for the case of k = 250 md, there is almost no improvement in production due to surging; for the case of k = 2.5 md, however, the boost is quite appreciable. This has an analogy in hydraulic fracturing, where greater fracture iengths are required in low perm formations.

Coals Which Display More Soil-Like Behavior In some well files of cavity operations in the faimay, the description sounds very much like soil behavio~ e.g., “coals are running; “massive coal returns during cleanout” etc. Furthermore, prolonged cleanout periods, characterized by massive -.._——— coai returns, are not accompanied by any pressure surging. The

7

L PALMER, H. VAZIRI, M. KHODAVERDIAN, J. MCLENNAN, K. PRASAD, P. ‘EDWARDS, C. BRACKIN, M. KUTAS, R. FINCHER

SPE 30012

—

--

Tdk

30

. ..*...

Phak 1

1 ~

<

\

..

\

%

40

\

\

,.. = ,.

30

10 % ●’

,..

.. .

..

.. . .

/.

120

‘>’ ,..

Km

,.’”

\

20

,.

.*

,.

,.

which is more soil-like.2 in fact, Figure 14 of ref. 2 illustrates this cavity and shear failure radii (and flow rate) are plotted against cohesion. The lowest cohesion is a factor of 20 iess than the cohesion .. .. used in an initiai match of the COAL site data. As a result, me cavity radius is increased from about 5 ft to i 2 ft, and the shear faiiure radius is increased from 35 ft to 60 ft. Aiso, fiow rate is increased by more than a factor of 2. This earlier study appears to illustrate more soil-like behavior of coals in certain parts of the fairway. That is, in such wells the cavity maybe sub stantiaily larger (24-ft diameter) than the 10-ft diameter at the COAL site, and the shear failure radius may be substantially iarger also. One interesting speculation: In these soil-like coals, the failure may be seif-sustaining in that, for low cohesion coal, there is a iarge shear failure zone in which the cohesion could be reduced even further by the sheer faiiure.

~

.“

~.: ,---

80 ‘-

.. a“” ,.~

60

40

20

0

0

0 10

1

~“

100O

100

i

Figure 11. Dependency of tensile failure (cavity radius), plastic failure (shear failure zone radius), and flow rate on perrnesblllty. —k=25ud 40

1“”

,

-I

k=23md

........

1

1

:,

b

k= 250M6 J

,

1 w

,

070

t m

,

lM”;

, m

, m

J an

mm ohm lM&oo (bun) Figure 14. One set of natural surges measured at both csvlty and obeervatlon well.

I..,,1,

-120

o

1

,,,

1

l,,!

1

2

3

1

1

4 T-

s

6

1

i

7

8

(lx)

Figure 12. Varlatlon of flow rate with permeability. 6

If successful cavity completions in the fairway impiy small-scale intense fracturing, in the baaai seams especiaiiy, we can use this information to predict iocations where cavities outperform hydrauiic fractures. Shce produtiion can be reprs sented as a convolution of the following principal factors: Production = Perm * Thickness * Pressure * Completion

1

Also, since the cavity development becomes iess, we may write:

improves as cohesion

Production = (Perm/Cohesion) * Thickness * Pressure

i

4

The key to coalbed methane success in the fainvay maybe in. hinh the term Perm/Cohesirm. ~- .. . ..=.. oualitv. ~-—.., , . . . . . —----- .- ... which ~h~~ld be l~rae intensely fractured basal coais. Conversely, a high Perm/Coh& sion suggests normal fracturing procedures maybe inadequate in the faitway, because (1) to stimuiate high-perm formations effectively requires a short, fat fracture (Frac-Pack or Tip Screen-Out type), (2) intensely fractured coais impiy ubiquitous coai fines, which can plug a proppant pack and destroy the fracture conductivity. 1 [ I

,

1

10

100

. —....,> ~

1

Suddenness of Blowdowns

moo

.—m

Uw

Figure 13. Increase In flow rata between drilling and end of surging as a furkctlon of Initial perm. hole is unstable, and the drilling gets sticky due to shale ledges and/or coal fines exiting a cavity into the wellbore. All this seems to imply that coal is continuing to fail under a hydrostatic head of liquid in the wellbore (possibly an underbalanced condtiion since fairway wells were often overpressured initiafly).1 This kind of behavior has been modeled previously, using a failure surface

589

A surface pressure valve is opened very quickly, allowing the pressure to fail to atmospheric within just a few seconds. However, the pressure at the bottom of the well wiii fall more slowfy, probably taking 1-2 min to faii to its equilibrium flowing pressure. Our model, which assumes the bottomhole pressure tracks the surface pressure with no time deiay, therefore overestimates the suddenness of the pressure surges. Consequently, the beet match cohesion should bean upper iimit (i.e., for slower pressure changes at the bottom of the wail, the cohesion would have to be lower in order to achieve the same cavity dimension).

.

.

COMPLETIONS AND STIMULATIONS FOR COALBED METHANE WELLS

8

Reoavitated Wells Several wells have been recavitated a few years after the original cavity operation, using a technique patented by Amoco.s The reasons for this included (1) the well was underperforrning compared with offset wells, (2) to clean out fill in the well, (3) the well was not cavitated for a long enough time originally, and (4) to replace tubing with larger diameter tubing. As the stabilized rates show in Table 1, these recavitations performed by Amoco have been highly successful (note that the wells were producing in the range 100+6,700 MCFD before recavitating). This euccess can be attributed to ●

cleanout of fill in the well

●

reduced friction within larger production tubing improving the original cavity complWlon5 despite substantially lower reservoir pressure at the time of recavitation

●

Pulse Analysis During Cavity Opemtions A well in the fairway of the San Juan Basin was cavity-completed in 1991 and the well was recavitated in 1993. An old Pictured Cliffs well, 242 ft from the cavity well, served as an observation well, and measured pressure interference data. Two sets of cavity surgings were done four days apart, and during both surgings, a rapid response was seen at the observation well. Pulse peak delays were about 16 min (see pulse delays shown in Figure 14). As well as the analysis of the overali pressure interference, a pulse analysis was applied to both sets of surgings.11

Tabie 1. Reoavitated Weiis

Recav #6

1.24 1.44

Recav #8

I

2.60

Recav #1 O

1.85

Recav#11

1.39

Recav #12

7.83

Recav#13

1.57

Average

2.27

I

b. Increased coal failure by surging: results from cavity modeling at the COAL site suggest between initial drilling and the end of six pressure surges, the avera~e cam (between cavity and observation weii 242 ft a~ay) incr&es by a f~or of 1.7.3 This agrees very well with the increase in interwell perm obsewed here between pulse set 1 and pulse set 2 (a factor of 1.74-1 .96). 5. The pulse technique also provides a new way of determining stress/dependent permeability. In every case, the blowdown perms are iower than the buildup perms, presumably because the pore pressure is lower, and the effective stress is higher. However, a strese-permeability function that is stronger than that normally quoted for the San Juan Basin is needed to reproduce this discrepancy.6 6. The fact that perrns measured during buildup exceed those measured during drawdown implies normal stress4ependent permeability, and means that the reservoir pressure of 700-750 psi is not yet low enough that matrix shrinkage dominates.

Comparison of Fraoture Treatments in the San Juan Basin A comparison was made between the various fracture treatments tried on dry coais south of the fairway (Figure 1) in the San Juan Basin. This work buiids on previous comparative studies.’ The coals in this area are typically about 3,000 ft deep, with permeabilities 0.1-1 md and net mal thickness averaging 50 ft.

The main findings of this analysis areas follows: 1.

4. When a comparison is made between pulse set 1 and puise set 2 (either DIowdown or biiilditip vaiues), the inWrwell perm increases by a factor of 1.74-1.96. This can be explained by

A similar pulse analysis has been done at the COAL site in the northwestern pari of the fairway (Figure 1). An observation well, 178 ft away from a cavity well, saw a direct pressure response during six injection/biowdown surgings on Day 8.1113114117 We emphasized the second half of the pulses, as they were more consistent. Our best estimates of time delays varied from 7-16 min. Although there are more uncertainties in thie analysis, we favor an interpretation in which interwell permeabilitiea lie in the range of 6-32 md, which a rees more with inde pendent results that give higher perms13’14r~ 5 than with the result that gave lower perrns18 at the COAL site. Once again, the permeability discrepancies between injection and blowdown reflect a strong stressdependent permeability.

1.47

Recav #9

3. Under non-blowdown conditions, the perm from pulse set 1 agrees with the perm from the preavity pressure match.

a. Drillbit access and cleanout of the bottom seam, which did not cccur until after pulse set 1, or

The last could be due, for example, to lower cohesion due to lower effective stress or lower water saturation or C02 desorption over time, or it could be due to removal of a low perm barrier by fines movement. Unfortunear $hn,“ -eqfitu ,., vQ@Ifie. ,.---, ea~sed -nately, separating the last effect from the first two is difficult, and has not yet been done.

Recav #7

2. Pressure pulses are seen at an obsewation well 242 ft away from the cavity well as a result of “natural” surges at the cavity well (i.e., no air or water injections, only shut-ins followed by blowdowns). AWwater injections are therefore not needed to see a direct pressure response at an observation well 242 ft away. This would seem to imply matrix communication, rather than fracture mmmunication (as hypothesized before).l

The pulse technique is a new application (of an old technique) to calcuiate permeability near a well during openhole cavity operations.

590

The database for the comparison consisted of fracture treatment data from approximately 1,000 wells located south of the fainvay. Several different operators and service companies were represented in the database over a wide geographical area. The wells in the database had production dates ranrairw from April 1989 to September 1993. h&et wells had produ~ofi

*

refxxtad for at least 12 months from the fracturing date and several had production data for up to 34 months. Monthly gas production data were obtained on these wells to evaluate the relative performance of various treatments. For this study, treatment details such as total fluid and sand volumes, pump rates, type of sand, etc., were not considered. The wells were grouped into three broad fracture treatment categories . Slick water fractures (friction reducer added to water). .

Similar comparisons have been carried out north of the fairwav in the San Juan Basin. and these results are shown in Table 3.

Table 3. Comparison of Fraoture Treatments North of the Fairway ) I

[SW) 1

. Croes-lrnked gel fractures.

Foam/SW

XLG/SW

Ratio of av. production over 6 mos.

1.24

1.06

Ratio of av. production over life of wells

1.41

0.94

11

132

Ratio Definition

For comparison purposes, localized areas with interspersed wells subjected to different fracture treatments were chosen. This was done to minimize the impact of geological variations on the fracture treatment response. Two production variables were used to evaluate the efficiency of treatments (1) average production over first six months from the treatment date, and (2) average production based on total cumulative production to date (i.e., average over life of wells). The results of the comparisons are summarized in Table 2.

Table 2. Comparison of Fracture Treatments e-. .+hAcE*:-.. -.. =UULII UI ruuwuy

#of wells These data show that ●

●

Water

(W’)

Foam M* Linear *, (C02LQ

va. Sllck Water

Cross-linked Gel (XLG) va. Foam c#!&kd Gel (XLGF)

Foam with

Llneer Gal(LQF) vs. Sllok

Water

Ratio of av.

C02LW

SW —..

XLGIXLGF

LGFISW

1.07

1.04

1.04

0.65

1.20

1.05

1.07

0.77

N@W

Water Fracture Treatments With and Without Sand in Warrior Basin In tha Wrsrrinr Rndn in Alnhcirna Ic$tu,

wells #of wells

48

27

25

18

.. s

The following can be concluded from these dati. ●

●

Nitrogen foam fractures without gel are better than slick water fractures over the life of the wells. However, there is Iiffle difference when considering only the first six months of production. There is no significant difference between C02 foam fractures and slick water fractures.

s There is no significant difference between foam fractures with cross-linked gel and liquid cross-linked gel fractures. ●

Water Fraoture Versus Foam Fracture Treatments in Arkoma Basin

●

produdon over 6 rnos. Ratioof av. -on over Meof

Overail, wells tilmulated using cross-linked gel do not perform as weli as those stimulated with slick water treatments. The ratios in Table 3 are inflated because of one anomaious area with significantly better cross-linked gei performance.

In southeast Okiahoma, in the Arkoma Basin, the Hartshorne coai (6-12 ft net) has been exploited by a number of operators in recent years. Evidence is that water fractures outperform foam fractures with gei.4 it has been hypothesized that in these coais, which are fairly dry, the foam bubbiee do not break easiiy, :-A- kAI-A J311=L AU-A III 1- tl[iuwlg I:-:*: -.. gutr --- cl-... -“AIU uhA.,. Iur= :w um 4---IUUIII urut.m IIuw mu mu ar hydraulic fracture.4 The observations are reminiscent of a group of coalbed methane welis stimulated ahead of mining in the eastern U.S.A.: anecdotal evidence is that water fractures again outperformed all other fracture treatments, including foam with gel. Note that this result, for fairly dry coals (typicaiiy ~0 bwpd), is the same as that for the dry coals south of the fairway.

(W)

(W) Ratio

Wells stimulated with foam fracturing treatment perform much hatter etimiIldari with elielr ● es., , ,e, ,,. --.. -, than ., ,S, , thnea ., ,-” “.,, ,,“, W...,., , “,, ”,. wdar .. s.”, trnatmant

co~ Nitrogen Foam :~os.

Cross-1inked Gei (XLG) VS. Siick Water (SW)

Foam vs. Siick Water

Foam fractures - these included (1) N2 foam fractures without gel, (2) N2 foam fractures with linear gel, (3) C02 foam fractures with linear gel,and (4) foam fractures with cross-linked gel.

Dafinltlon

9

1.PALMER, H. VAZIRI, M. KHODAVERDIAN, J. MCLENNAN, K. PRASAD, P. EDWARDS, C. BRACKIN, M. KUTAS, R. FINCHER

SPE 30012

Slick water fractures outperform foam fractures with linear gel.

591

w .“

..-s

..”,

“=”,,

v

*8 v ru9ua,

,,a,

im ,, ,

+ha u ,9

nab wan

~.mt,a UIVVU

CI.4A 1

three principal coal groups are produced: the Pratt, the Mary Lee/Blue Creek, and the Black Creek.7 In the western side of the Oak Grove Field, near the Black Warrior River, typicai depths are 1,200 for the Pratt, 1,600 ft for the Mary LeeiBiue Creek, and 1,900 ft for the Black Creek. Note that the Pratt and the Mary Lee probably provide almost 90% of the total gas from the three coal zones. --:-

IIw

Recent AmcrcWaurus fracture treatments were of two --A_ ...!*L -_—J -—J ——— J,-. — &----- .... .. . Z.-A. --- A---A_ wuLur iramuru mrimmerm wwr aana, ana sanawes

t iy~i

water fracture treatments (the latter used baii seaiers to try to open up more pads). Piain water with no additives was used in both cases. In typical water fracture treatments with aand we wmrmd around 70.000 lb of 12/20 sand at concentrations &6 I& at rates of 4&60 bpm. The eandleee water fracture treat-

.

10

COMPLETIONS AND STIMULATIONS FOR COALBED METHANE WELLS

rnents were pwnped at around 60 bpm using ball sealers of SG 1.3 (about 150 balls per zone)? The location of the wells used in this comparison is shown in Figure 15. There are six wells with sendlesa water fracture treatments and eleven wells with water fracture treatments with sand. Some wells further away from the river have undersaturated coals. Because these tend to produce very high water rates, and very low gas rates, they have been excluded from the above comparisons. In only one case was the Pratt seam not f,ed, ,Pn tra tmant) m t$ia ehnt M nnt df6rt ~:m, ,B~*4 I=LIIIIUI=CUU [a ..,eb-~ waLw,,,~“,”, = Uau.gv...,, -.. .“ “..--... ..V. -..-4. the conclusions. Although the distribution of the wells in Figure 15 is not as random as might be desired to rule out any geological variability, we feel that the distribution is random enough to make a valid mmparison, The difference between the two frac types is shown in Figure i 6, and for gas prcdutiion it is clear that the water fracture treatments with aand outperform the aandlass water fracture treatments up to a time of 41 months. The big jump in water rates just after 40 months in Figure 16 is due tQ Qnly a few wells remaining on line after this time, so these data are invalid for comparison. At a time of 41 months after production began, the cum gas production from the water fractured wails exceeds those from the sandless water fractured wells by 1.40:1. This can be compared with gas rates that differed b about a factor of 1.8, in the same dirtilon, reported previously. #

give less water production. This is difficult to rationalize, but may reflect the wide variability in water production (there is much less variability in gas production). Based on a simplistic Calculation,g it can be shown that economically the water fractures with sand, although more expensive, are still better than those without sand, since the difference in average rates (at 41 months) implies a payout time of 76 days. Possible biases in the data include

●

●

●

6/11 of the wells with water fractures with sand were subject to a remedai water-bailout treatment some time later. in several cases this remedal treatment appeared to increase significantly the gas praiuction. The weiis are not entireiy interspersed, but we can wrnpare the two fracture treatment types after removing (1) the group of four water fractures with sand in the iower left of Figure 15, and (2) the three sandiess water fracture treatments in the unoer -rr -. right of Flaure 15, and this gives the resuit in Figure 17. The discrepan~ in gas production is a little greater, but the conclusion remains the same.

—

— -ml~

Figure 16. Discrepancies in gas and water production

—

— .

I

.

—

Figure 17. Same as Figure 16, except outlying weiis are removed. Figure 15. Interspersed wells used to compare different fracture treatments.

Foamed Water Cieanoute in San Juan Basin

The data indicate that while the water fractures with sand give greater gas production than sandless water fractures, they

592

Foamed water cieanouts (FWCOS), a technique patented by Amoco,8 were performed on 19 coalbed methane WOM in thO

. SPE 30012

L PALMER, H. VAZIRI, M. KHODAVERDIAN, J. MCLENNAN, K. PRASAD, P. EDWARDS, C. BRACKIN, M. KUTAS, R. FINCHER

San Juan Basin several years ago. Three treatments were initial completions, and 17 were remedial treatments following cross-linked gel fracture treatments. Fifteen of the wells treated were below average producers, compared to offset wells. Only one of the three initial completions was commercially successful. This is probably because at typical depths in the basin (1,500-3,000 ft) the in-situ stress will close a fracture that is not propped open. The FWCOS were conducted down casing using mostly 70 Q foam pumped at 10-40 bpm. Treatment volumes varied from 3-6 bbl foam/ft of coal. Each treatment was allowed to flow back immediately to the pit, unrestricted, until clean returns were obsemd. Generally the treatment was repeated at each well. Results: About 759f0of these wells experienced increases in gas production, although only about 50% were economical (based on $1.50 per MCF gas price and six-month payout time). Treatments were unsuccessful on two wells that were better than average. Average ratio of postproductiordpreprcduction is 3.0. However, two ratios were much higher than all others (6.2 and 14), and when these are ignored, the average falls to 1.8. Note that initial gas production varied from 10 MCFD to 170 MCFD. Total gas and water production increased by approximately 1,100 MCFD and 600 BPD, or by 629’o and 23% respectively. Note, however, that three FWCOS on two wells account for more than half the incremen@l gas rate increase.

These treatments should not replace sand fracture treatments for initial completions. However, in many cases they do boost production rates significantly as remedial treatments. Since sand is returned during the flowback portion (plus some coal fines), our interpretation is that the FWCO treatments are creating new flow channels in the proppant pack, and thereby increasing the fracture conductivity.

Hydraulic Fracture Geometry This subject was summarized previously.7 Based on extensive observations in both San Juan and Warrior baains, fracture treatments are predominantly of two types

●

vertical fracture strands (each with sand) diverged over a wide angle, seems less likely than the T-shaped fracture geometry. However, unpublished information from water fracture stimulations ahead of mining in an eastern coal field reveal that horizontal fracture components were very rare, and only occurred if the fracturing gradienta (i.e., ISIP) exceeded 1.8 psi/ft. All these wells had very poor performance. The majority of the hydraulic fractures were vertical, extended up to very large distances from the wellbore (half-fengths of 800 ft or more), and some had fracture gradients in excess of 1.0 psi/ft. In the water fracture treatments, sand stages were separated by pad stages, and presumably the pad stages sweep the sand away from the wellbore as the sand banks along the bottom of the fracture. Finally, some mineback statistics from eastern coalbeds of ten fracture treatments with bottomhole pressure >1.0 psi/ft, only five had horizontal fracture components.’ Furthermore, these five were not all >1.8 psi/ft, so the 1.8 psi/ft criterion given above for a horizontal fracture component may only be true for that specific area.

Concluding Remarks 1. Field data on cavity completions: Flow rate increases during cavity operations vary a lot, and can be as high as 8:1. +alv In nnn T1.a -Lb-.-.da. A{ +ha fkm .mta in-r zma vmrlae I I lu u Ial auLOl WI u IG Ilww I U*9 !1W%”.-! !. !“”!,. . . ! “! 0“ area, pressure surging was applied for seven days before significant coal failure. In other cases, flow rate increases markedly, without pressure surghg, during cleanout operations (i.e., the coal is continually failing). Based on nine cases, bottomhole pressure buildups after one hour in natu-

lnj~lon rates of 2040 bpm were the most successful. Results were best when treated wells were flowed back immediately, and at maximum possible rates. One well in the fairway was treatad twice, each with three cycles, and production results were excellent, increasing from 55 to 370 MCFD.

●

11

High pressure treatments in which bottomhole pressures are flat or rising, and ISIP exceeds 1.0 psi/ft. These are fractures confined to a coal seam, and are likely to be T-shaped geometries. Bottomhole pressures that fall with time, with ISIP less than 1.0 psi/ft. These reflect fractures with height growth into bounding zones of lower stress.

Examples of both kinds have been found in minebacks in the Warrior Basin, as reviewad elsewhere.7 No minebacks have been conducted in the San Juan Basin. However, in four out of six cavity recompletion in the fairway of the San Juan Basin, frac sand was returned to the surface during cleanout periods. These were cases in which the original well had been hydraulically fractured with sand, and the well later sidetracked to allow the cavity recompletion - the sidetrack is typically only about 30 ft from the original well. Now if the original fracture were a vertiial fracture, the chances of a typical 10-ft diameter cavity intercepting such a fracture would be far less than 4/6. The simplest interpretation ~. —-—._ of the observation is that the original fracture was T-shaped, and the horizontal component would have intercepted the new CSv~ weii. The o~iy 0ihi3TpOS3ibiitiy, thtii ~ Fiufi&f Of

593

red ““, RI wminrie gwsramn ,-, ~.. l=” “.”,-=”

7A nsi/ft --y-” ... Fnr . -. ~~~ iniactirm ..... . ... . . surainas. --. =...=-,

the average is .45 psi/ft. In all cases, air injection pressures are less than .59 psi/ft, which is lower than any virgin horizontal stresses measured in basal seams. This should imply injection below parting pressure, but the parthg pressure itself may be lowered due to stress relief around the wellbore.2 For air and water injections, bottomhole injection pressures are .80-1.14 psi/ft, and these all exceed the average minimum stress in the basal seams, implying fracture injection. 2. Cavity modeling results have been summarized. A principal benefit of the modeling has been to provide values for large-scale cohesion. The cavity stabilizes fairly quickly in the model, but often not so in the field, perhaps reflecting either lower cohesion coal (i.e., more soil-like) or coal heterogeneity. a. ~dt!~~l ~rarneters for su~essful cavitation: Cohesion is the most critical parameter. However, another important parameter is the pore pressure gradient near the surface of the cavity. This is a consequence of (a) Iarnehpid pressure changes at the wellhead, and (b) formation permeability. A third parameter, the in-situ stress condition, has less importance. A “stress threshold’ effect on cavity performance is not seen in the modeling. Finally, very low cohesion (i.e., soil-like behavior) probably explains cases in which coals are “running; and accompanied by “massive” coal returns at the surface, sometimes in the absence of pressure surging. This suggests cavities in excess of the typical 10-ft diameter measured in the field. 4. The key to coalbed methane success in the fairway maybe the ratio of perm/cohesion. Where this is large, cavity comnidinns.- ShOUI~ work better than fr~dure stimulations, and ~----vice-versa. 5. Recavitations have been very successful, but it is difficult to determine whether cleanout of fill in the well accounts for

COMPLETIONS AND STIMULATIONS FOR COALBED METHANE WELLS

12

this alone, or whether the original cavity completion is actually improved. 6.

7.

Pressure interference has been obsewed on a very shorl time scale at an observation well near a well being cavitated on two occasions. Air/water injections are not needed to see direct pressure response at the observation well, implying matrix communication and not fracture communication. A pulse interference analysis is used to calculate interwell permeekility, which largely agrees with other measurements. The technique appears to provide a new way of determining stressdependent permeability. Comparison of fracture treatments in the San Juan Basin: (1) South of the fairway, nitrogen foam fractures with no gel outperform slick water fractures. Slick water fractures outperform foam fractures with linear gel. (2) North of the fairway, wells stimulated with foam fracturing treatment perform much better than those tiimulated with slick water treatment. Overall, wells stimulated using cross-iinked gei do not perform as well as those stimulated with slick water treatments. The ratios in Table 3 are inflated because of one anomalous area with significantly better cross-linked

2.

Palmer, 1. D. and Vaziri, H., “Modeling of Openhole Cavity Completions in Coaibed Methane W3ii%” W% 28580, Procs. SPE Ann. Tech. Conf., New Orleans, LA, September 1994.

3.

Khodaverdian, M., Vaziri, H. Palmer, l., McLennan, J., “Openhole Cavity Completions in Coalbed Methane Wells Modeling of Field Data; Procs. Intergas ’95 Intl. Uncon. Gas Symp., Tuscaloosa, AL, May 1995.

4a.

Penny, G. S. and Conway, M. W., “Coordinated Studies in Support of Hydraulic Fracturing of Coalbed Methane: Annual Report to GRI, GR1-94/0398, August 1994.

4b.

Conway, M. W., Smith, K., Thomas, T., Schraufnagel, R. A., “The Effect of Surface Active Agents on the Relative Permeability of Brine and Gas in Porous Media; SPE 28982, Procs. SPE Intl. Symp. Oilfield Chemistry, San Antonio, TX, February 1995. Method for Fa!mer, i. ~. am~ gdw~r~e .P . ,A... l~rA~ irnnr~vd ~----.-. ,,” b“...-.-”,

5.

Stimulating a Coal Seam to Enhance the Recovery of Applic. Methane from the Coal Seam; Pat. No. 06/250,561, Pending, May 27, 1994.

flel ~-. oerf~rm~n~e, ~-

&

8.

Water fractures outperform foam fracture treatments with gel in the Arkoma Basin, and apparently in one play in the northeastern U.S.A. This is the same result as south of the fainvay.

s~idl~, J, P,, Jeanaonne, M. J., Erickson, D. J., “Application of Matchstick Geometry to Stress-Dependent Permeability in Coals; SPE 24361, Proos. SPE Rocky Mtn. Reg., p. 433, Casper, WY, 1992.

7.

9.

In the Warrior Basin, water fractures with sand outperform sandless water fractures (in gas prd”utiion) by a factor of

Palmer, L D., Lambert, S. W., Spitter, J. L., “Coalbed Methane Well Completions and Stimulations,” in Hydrocarbons AADR T, ,Ien 1 QO!l 4---e--l \w. l-d Law 1a.., allu en~ 1qlfi~\ II UIII Uual, ,“-,, m, w, , “,”-, .“”-.

n “.

k, it..

1, .-r, A Q,,“ zind +hau ● ,“,

10.

sw-“ hdtar “, ““..”. ammnqidlv -- ... .. .. . ... .

Foamed water cleanouts in the San Juan Basin boost gas production rates significantly as remedial treatments. The treatments probably create new flow channels in the proppant pack.

11. Hvdraulic fracture mometw: In the San Juan Basin there is new support for T-_haped _fracture geometry. On the other hand, in a play in the northeastern U.S.A., apparently most fractures were vertical, even when frac gradients exceeded 1.0 psi/ft, and horizontal fracture components were very rare. Acknowiedgniems We are grateful to Don Walton of Halliburton for providing assistance in compiling public date that was used to compare different fracture types south of the fairway in the San Juan Basin. Dale Kalcich of Halliburton assisted in compiling the database for fracture comparisons north of the fairway in the San Juan Basin. X. Wang of Technical University of Nova Scotia is acknowledged for performing the numerical analyses of the cavity model. Kent Sauvaugeau, Rudy Candelaria and Pete McNeal of Farmington O.C. are thanked for their support and assistance in acquiring the data for the pulse analysis. Reggie Waiters, Don Morgan, and L. G. Huitt of AITC are thanked for their spreadsheet and plotting analysis. Margie Meyer has been excellent in her desktop publishing support. Gas Research Institute, and Dick Schraufnagei, are acknowledged for their support of a contract to TerraTek under which part of this work was done. They also sup ported, along w“th Amoco, the original research at the COAL site. We thank Amoco for permission to publish this paper.

594

Wamariisl ~~~~~rn~n?for’ ~Qai Deg~S We!!%” M w. ,.,., ! ,“. ..-.-.

C

Pat. No. 4,913,237, April 3,1990. 9.

Palmer, L D., Kinard, C. M., Fryar, R. T., “Sandlees Water Fracture Treatments in Warrior Basin Coalbed%” Procs. 1993 Intl. Coalbed Methane Symp., Birmingham, AL, May 1993.

10. Young, G. B. C., Kelso, B., Paul, G., “Understanding Cavity Well Performance; SPE 26579, Procs. Ann. Tech. Conf., New Orleans, LA, September 1994. 11. Khodaverdian, M. F., McLennan, J. D., Palmer, L D., Vaziri, H. H., “Spalling and the Development of a Hydraulic Frac!~~ng stra?eg)/ for coal: Final ReDort to GRI, 1995. 12.

Muthukumarappan, R., “Analysis of the Success of Openhole Cavity Completions in the Fairway Zone of the San Juan Basin: Ph.D. Thesis, Mississippi State, MS, December 1994.

13. Mavor, M. J., “Coal Gas Resewoir Cavity Completion Well Performance; Procs. 1992 Intl. Gas Research Conf., Orlando, FL, 1992. 14. Logan, T. L., Mavor, M. J., Khodaverdian, M., “Optimization and Evaluation of Openhole Cavity Completion Techniques for Coalbed Methanev SPE 25859, Procs. Rocky Mountain Regional/Low Perm Sym., Denver, CO, 1993. 15. Young, G. B. C., Kelso, B. F., Paul, G. W., “Understanding Cavity Well Performance; SPE 28579, Proca. Ann. Tech. Conf., New Orleans, LA, 1994. 16.

Jochen, V. A. and Lee, W. J., “Reservoir Characterization of an Openhole Cavity Completion Using Production and Well Test Data Analysis; SPE 26917, Procs. Eastern Reg. Conf., Pittsburgh, PA, 1993.

17.

Mavor, M. J. and Logan, T. L., “Recent Advances in Coal Gas-Well Openhole Well Completion Technology” JPT, p. 587, July 1994.

References 1. Palmer, L D., Mavor, M. J., Seidle, J. P., Spitter, J. L., Volz, R. F., “Openhole Cavity Completions in Coalbed Methane Wells,” JPT, Vol. 45, p. 1072, November 1993.

,,”.e”,

1.PALMER, H. VAZIRI, M. KHODAVERDIAN, J. MCLENNAN, K. PRASAD, P. EDWARDS, C. BRACKIN, M. KUTAS, R. FINCHER 18. Soott, A. R. and Kaiser, W. R., “On the Geomeohanioe of Open-hole Cavity Completions in San Juan Basin Coalbed Gas Wefts,” Pres. Nofih American Rook Mechanics Symp., University of Texas at Austin, TX, June 1994. 19. Mavor, M. J., pars. oomm., 1994.

595

13