Formation Evaluation of Exploration Coalbedćmethane Wells
Formation Evaluation of Exploration Coalbedćmethane Wells
Formation Evaluation of Exploration Coalbedćmethane Wells
CoalbedĆMethane Wells
M.J. Mavor,* SPE, and J.C. Close,** SPE, Resource Enterprises Inc., and
R.A. McBane, Gas Research Inst.
Summary Powder River, Hanna, and Green River16,17 (Wyoming); and Sand
Development of coalbed-methane reservoirs requires accurate de- Wash18 (Colorado and Wyoming) basins. Geologic setting, maceral
termination of gas in place and natural fracture permeability. This and chemical composition, coal rank, and tonnage estimates for Ca-
paper outlines steps needed to evaluate reservoir properties and to nadian coals have been discussed in detail.19-22 The major lack of
predict well deliverability. Procedures are documented with data accuracy in resource estimates results from uncertainties concern-
and analyses from San Juan basin development projects. The paper ing reservoir gas content and net pay thickness. In developed basins,
includes an example of the accuracy of gas-content determinations such as the San Juan and Black Warrior, resource estimates are suffi-
from pressure-, sidewall-, and conventional-core and drill-cuttings ciently accurate to help estimate development economics. In unde-
samples; presents specifications for complete log suites and analy- veloped basins, gas contents are unknown and must be quantified
sis techniques for gas-content estimates; discusses drillstem-test during the initial drilling program.
(DST) procedures; and outlines prediction of well deliverability. Coal structure, stratigraphy, and thicknesses are relatively well
known because of numerous geologic studies. The gross thickness
Introduction and ash (noncoal) content of coal zones can be determined from well
logs of wells drilled to deeper horizons. In most U.S. basins, a signif-
This paper presents examples of the method required to evaluate the icant amount of well control exists. Unfortunately, gas content can-
properties of coalbed-methane reservoirs for deliverability predic- not be determined from these logs until gas-content data are avail-
tion. Accurate projections of fluid production rates over the life of able in the coal seams of interest. Permeability is generally unknown
a reservoir are necessary to make informed decisions concerning the before drilling except at a few specific locations that have been dis-
economic viability of coalbed-methane reservoir development. cussed in the literature. The existence of permeability can be deter-
Fluid-flow-rate projection requires estimates of reservoir properties mined when micrologs, drilling-mud densities, or production data
that are components of Darcy’s and Fick’s laws, such as reservoir are publicly available. Ref. 23 discusses details concerning proce-
geometry, fluid volumes in place, fluid properties that affect fluid dures for qualitative estimation of Fruitland formation coal-gas res-
flow, and rock properties that affect fluid flow. ervoir permeability before drilling.
We used a four-step process to determine the properties of coal- Proper collection of accurate data is of prime importance in the
gas reservoirs. Step 1 is to review available, published information exploration and early development stages for a reservoir. Much of
in the geologic and petroleum engineering literature concerning the data required for deliverability prediction can be collected only
coal properties in the geographic area or geologic formation of inter- at this time. The following section discusses collection of coal sam-
est. Step 2 is to collect core, well-test, and wireline log data from ples, required core analyses, wireline-log data interpretation, and
newly drilled wells. Step 3 is to perform laboratory core, log, and openhole DST data collection and interpretation.
well-test analyses to quantify reservoir properties estimates. Step 4
is to combine the data and to predict future fluid production rates
with a coal-gas reservoir simulator. Sample Collection and GasĆContent Estimates
The remainder of this paper discusses the collection and evalua- The primary objective for collecting coal samples is to estimate the
tion of these data. Procedures are illustrated by examples from three in-situ gas content (gas volume/rock mass ratio). At the present
wells drilled in the Fruitland formation coal-gas reservoirs, San time, accurate estimates of coal-gas content require wellsite mea-
Juan basin, Colorado and New Mexico. The wells were cooperative surement of the volume of gas liberated from coal samples. Types
research efforts of the Western Cretaceous Coal Seam Project, of coal samples that can be collected include shale-shaker drill cut-
funded by the Gas Research Inst. (GRI), and were performed by Re- tings, conventional cores, drilled sidewall cores, wireline-retriev-
source Enterprises Inc. and the cooperative operators. McKenzie able cores, and pressure cores. Other major reasons for collecting
Methane Co. operated the first well, Southern Ute–Mobil 36-1,1 in coal samples are for gas storage capacity measurements, natural
Section 36, Township 34 North, Range 10 West, La Plata County, fracture geometry description, laboratory flow experiments, and to
CO; Mesa Ltd. Partnership operated the second well, Hamilton No. determine coal rank and composition.
3,2 in Section 30, Township 32 North, Range 10 West, San Juan Gas content and rate of gas desorption from coal are determined
County, NM; and Blackwood & Nichols Co. Ltd operated the third by canister desorption experiments. The total gas volume originally
well, Northeast Blanco Unit No. 403,3 in Section 9, Township 30 contained in the coal is the sum of the lost, measured, and residual
North, Range 7 West, Rio Arriba County, NM. gas volumes. Total gas content is total volume of gas divided by
mass of the sample. Desorption experiments are performed by seal-
Predrilling Data Sources ing coal and other rocks in a canister and measuring the volume of
gas evolved from the sample as a function of time, temperature, and
General background information concerning potential coalbed-
pressure. The volume of gas evolved is measured by displacement
methane reservoirs abounds in the geologic and petroleum engi-
of water from a graduated cylinder. The major inaccuracy
neering literature for both coal mining and gas production. Informa-
associated with total gas-volume determination is the need to esti-
tion sources for U.S. basins include those for the U.S. in general4
mate the volume of gas lost while retrieving the samples to surface.
and the San Juan5-8 and Raton9-11 (Colorado and New Mexico); Pi-
Two methods are commonly available for this evaluation: the di-
ceance12 (Colorado); Black Warrior13,14 (Alabama and Mississ-
rect24 [U.S. Bureau of Mines (USBM)] and Smith and Williams25
ippi); northern Appalachian15 (Pennsylvania and West Virginia);
methods. Both techniques assume that diffusion from the coal oc-
*Now with Tesseract Corp.
curs from a spherical sample and is an isothermal process. The se-
**Now with Meridian Oil Inc. cond assumption is the most severe limitation. The direct method as-
Copyright 1994 Society of Petroleum Engineers
sumes that the external gas concentration of the coal sample is
immediately reduced to near zero. This is equivalent to assuming
Original SPE manuscript received for review Aug. 1, 1990. Revised manuscript received Aug.
10, 1993. Paper accepted for publication March 1, 1993. Paper (SPE 21589) first presented
that the pressure of the sample is immediately reduced from reser-
at the 1990 SPE/CIM Joint Intl. Technical Meeting held in Calgary, June 10–13. voir to atmospheric conditions. This assumption is not valid for coal
samples obtained while coring with mud. The Smith and Williams sorption experiments. ResidualĆgas content was negligible for all
technique attempts to account for the gradual pressure reduction but samples.
is limited by the fact that the decrease is assumed to be linear in time, Care should be taken when using Table 1 to compare the relative
which is usually not the case. The Smith and Williams method also accuracy of gas-content estimates. The estimated errors are for one
assumes that the samples are fully saturated with gas and that the 2,500-ft-deep reservoir under specific temperature and pressure
coal-gas storage capacity is a linear function of pressure between conditions. Error estimates may be low owing to the external tem-
reservoir and atmospheric conditions. Neither of these assumptions perature (*5°F) when the samples were retrieved to surface. The
is the usual case. temperature of the core barrel was approximately 45°F. It is likely
As a result of inaccuracies in the lost-gas-volume estimates, sig- that the cold weather caused lower lost-gas volumes than if the mud
nificant errors can exist in the determination of the gas-in-place vol- temperature had been greater.
ume. Pressure-core samples are not limited in this way because they The gas desorption rate is obtained by use of canister desorption
are sealed at reservoir pressure and retrieved to surface without loss experiments. The desorption rate is related to the diffusion charac-
of gas. Wireline-retrievable cores and mechanically drilled sidewall
teristics and size of the coal samples. The diffusion rate is propor-
cores can be placed in canisters within 15 minutes of coring, which
tional to temperature. Whenever possible, the desorption canisters
results in minimal lost-gas volumes. Drill-cuttings samples and
should be heated to reservoir temperature to estimate desorption rate
pressure-, conventional-, and mechanically drilled sidewall-core
samples were collected from Southern Ute–Mobil 36-1 well to and diffusion coefficients at reservoir conditions. At the present
compare the relative accuracy of gas-content estimates (see Table time, no accurate method is available to adjust the rate data to differ-
1). The estimated errors in this table were calculated assuming a ent temperature conditions. The quality of the data obtained from
95% statistical significance level. The mechanically cut sidewall- heated canisters is superior to that from unheated canisters, and the
core-sample gas-content estimate was within 5.6% of the pressure- residual-gas content is minimized.
core estimate owing to the short trip time. The sidewall-core gas- The pressure-core sample was collected from 2,428 to 2,438 ft
content estimate agreed with the pressure-core result when the with Eastman Christensen pressure-core equipment. Samples were
statistic error was considered. Because of the limitations in the lost- successfully sealed and retained reservoir pressure when retrieved
gas methods, the conventional core-sample gas-content estimates to surface and reheated to reservoir temperature. The entire core
were low by 7% to 13%. Drill-cuttings gas-content estimates were barrel was desorbed. After 9 hours of desorption, gas rates were suf-
low by 24% to 28%. Residual-gas-content estimates were deter- ficiently low to transfer the core samples to conventional canisters.
mined by crushing selected samples at the conclusion of the de- Long-term desorption volumes were added to the pressure-barrel
desorption volume to determine total gas volume. It was necessary the coal matrix and to escape as the pressure is reduced while the
to allocate the gas volume desorbed from the pressure-core barrel samples are circulated to the surface.
to each of the long-term desorption intervals on the basis of ash con-
tent. The average ash and total gas contents of the pressure-core in- Core Analyses
terval were determined to be 0.341 and 396.2 scf/ton, respectively.
Numerous analyses exist for characterizing coal samples. Table 2
The ash content is the fraction of noncoal material in the sample. lists the types, purpose, and relative importance of coal core mea-
The average ash-free total gas content (total gas content of pure surements. Analyses of primary importance are required to estimate
coal) was determined to be 601.4 scf/ton. gas and water production rates accurately. Analyses of secondary
Sidewall-core samples were collected from Well Southern Ute– importance are required for various purposes and are recommended
Mobil 36-1 with a Schlumberger mechanical sidewall coring tool. for all key well samples. Key wells are those that are cored early in
Significant gas volumes were desorbed and measured from the re- the life of the development program to obtain the data listed in Table
covered samples. Core recovery was less than desired owing to a 2. Studies to determine gas content and rank of the coal are recom-
combination of coal-mechanical competence and coring-tool de- mended for each coalbed-methane well. These analyses, each
sign. The ash content of the samples was greater than estimated from marked with an asterisk (*) in Table 2, would be performed on drill
the wireline logs, which suggests that low-ash-content coal was cuttings for routine wells and core samples as well as cuttings for
preferentially removed during the coring and/or retrieval process. key wells.
We inferred that this discrepancy did not result from an error in the Note that all analyses marked with a single asterisk in Table 2 and
log-derived estimates of ash content on the basis of agreement be- the sorption isotherm can be performed on drill-cuttings samples.
tween openhole bulk-density data, core computerized-axial-tomo- The major losses in accuracy are likely to be low total gas-content
graphy (CAT)-scan bulk-density data, and proximate analyses. estimates and lack of certainty in the depth of origin of the samples.
Excellent core recoveries have been obtained with conventional
plastic- or aluminum-lined core barrels. All well coring was per-
Gas and Ash Content
formed with 6.25- 3.5-in. conventional-core barrels. The time re-
quired to retrieve the core samples off bottom and to seal them in Proper interpretation of the gas-content data requires that proximate
canisters is typically 1.5 to 3 hours for Fruitland formation reser- analyses be performed. Approximate analysis26 is a standard coal
voirs from 2,500 to 3,000 ft deep. When the core reaches the surface, laboratory evaluation that determines the moisture, ash, volatile-
the handling personnel must be well coordinated to reduce the time matter, and fixed carbon content of the sample. These analyses are
required to seal the samples in canisters. Core-handling procedures performed on sample splits obtained from each canister at the con-
are as follows. The inner core barrel is removed from the coring clusion of the desorption experiments, which may be 4 to 6 weeks
equipment on the rig floor and laid down on the catwalk that is be- after coring. Sample splits can be selected and separated on location
low the V door. The core and liner are slid onto the catwalk from the to obtain needed data more promptly. Ultimate analyses,27 which
inner barrel. Core samples (still in the liner) are cut into 1-ft lengths determine amounts of carbon, hydrogen, nitrogen, and oxygen, are
with a rock saw. Each sample is pushed from its liner and placed in sometimes performed as well for selected samples.
an unsealed plastic bag, which is immediately sealed in a marked It has been possible to relate the gas content to the ash content of
canister. The canister is then taken to an on-site laboratory for the the core samples for all the northern San Juan basin research wells.
desorption experiments. Although bagged samples often retain suf- The ash content is most useful when converted to a dry basis with
ficient competence for flow experiments, the mechanical integrity the following relation.
of the samples can sometimes be improved if the samples are left in f ad + ań1 * w. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (1)
aluminum core sleeves or placed in plastic sleeves for transportation
to the laboratory. The relation between core gas and ash content can be expressed
Shale-shaker samples are collected in a large strainer and washed as follows for the three wells discussed here.
free of drilling mud before sealing in desorption canisters. Shale- G c + G af * cf ad . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (2)
shaker samples should always be collected for gas-content deter-
mination, even when a coring program is planned. Core recovery If all the gas were contained in the coal with no contribution from
sometimes can be poor, and the shaker samples may be the only reli- the noncoal material, constant c should be equal to the dry, ash-free
able source of gas-content estimates. gas content, Gaf . This was not the case, as is illustrated by the regres-
Major problems with drill-cuttings samples are depth control and sion results summarized in Table 3. (The estimated errors for each
gas-content estimate accuracy. When drilling thick seams or multi- of the parameters were calculated assuming a 95% statistical signifi-
ple adjacent seams, the origin of each sample can be difficult to de- cance level.) The predicted ash volumes for zero gas content were
termine, except in the case of the first samples obtained from the first 0.79 and 0.82 for the direct and the Smith and Williams analyses,
coal seam. The gas-content estimate from drill cuttings is typically respectively. We do not now have a proved explanation of why car-
lower than that obtained from other sources. Crushing the coal into bonaceous shale intervals with w80% ash content would be gas
particles the size of cuttings allows the gas to diffuse rapidly through free. One possibility is that the ash content reduces the effective dif-
fusion coefficient to a level where the gas is not released from the and the optical nature of the recognizable organic constituents in
coal during the desorption experiment. coal.29 The three primary maceral groups are vitrinite, exinite (lipti-
Table 3 also gives results from two other research project wells. nite), and inertinite. Each of these groups encompass subgroups
Estimated total gas contents from whole core samples were similar. with similar properties. In (U.S.) Upper Cretaceous Western Interior
The gas contents for Wells Hamilton No. 3 and Northeast Blanco coals, vitrinite is usually the most common maceral. Vitrinite mac-
Unit No. 403 are lower because trip times were longer than for Well erals are derived from the woody tissues of plant matter.
Southern Ute–Mobil 36-1. Long trip times were responsible for the
lower correlation coefficients for Well Hamilton No. 3. Note that Gas Storage Capacity. A sorption isotherm is a primary coal analy-
these gas/ash-content relations were developed for specific wells in sis that can be measured on whole-core or drill-cuttings samples.
a specific basin. The gas content as a function of ash content, mois- Isotherm data are required to relate the gas storage capacity of the
ture content, coal rank, and other coal characterization properties coal to pressure. The sorption isotherm is also a function of the
must be developed from core data in other basins or areas besides moisture content of the coal and the temperature of the experiment
the research wells. Isotherm data are obtained through adaptation of a Boyle’s law
porosimeter. A known gas volume (typically pure methane) is
Coal Rank and Composition. Coal rank is the assignment of a dis- introduced from a reference cell into a sample chamber, initially at
tinct maturation level to a coal derived through the measurement of atmospheric pressure, that contains approximately 100 g of coal at
chemical and physical properties. The American Soc. for Testing reservoir temperature and moisture content. The resulting pressure
Materials (ASTM) consortium devised a classification system that is used to compute the volume of gas that was adsorbed onto the
is based on the chemical properties of the whole coal and the optical coal. The experiment is continued by injecting additional volumes
properties of the vitrinite maceral group. The properties most com- of gas at progressively greater pressures until the range of the entire
monly used for rank classification in the U.S. include fixed carbon curve is obtained. Typically six to eight pressure steps are used that
content, volatile matter content, sulfur content, gross calorific val- range from atmospheric to a pressure exceeding initial reservoir
ue, and vitrinite reflectance. Coal-rank determination is important pressure. Both increasing and decreasing pressure sequences can be
because the capability of the coal to generate gas is related to the used to determine the sorption isotherm. For a single-component
rank of the coal. It has been suggested that significant volumes of
gas, in the absence of experimental errors, the two isotherm relation-
thermogenic gas are generated and stored when the Fruitland coal
ships are identical because the process is reversible under laboratory
rank exceeds 0.72% Ro .28 Coal-rank values are used primarily to se-
conditions.
lect locations and required drilling depths for coalbed-methane res-
Data usually are interpreted by assuming that they can be fit to the
ervoirs and to explain why measured gas contents may be less than
Langmuir30 relation. A Langmuir isotherm is represented as fol-
desired.
lows.
The same sample used to determine the vitrinite reflectance is
used to determine the maceral composition of the coal. The coal p
G s + V Lǒ1 * f adǓ . . . . . . . . . . . . . . . . . . . . . . . . . . (3)
maceral concept is used to describe the microscopic morphology p ) pL
VL , the “Langmuir volume,” is the maximum gas storage capacity orientation of the natural fracture systems and to obtain a qualitative
of the ash-free coal. pL , the “Langmuir pressure,” is equal to the understanding of the magnitude of cleat development (e.g., spacing,
pressure at which coal storage capacity is equal to one-half the height, aperture, and porosity). Polished-block fracture-analysis
Langmuir volume. Not all coals will ascribe to this simple relation- data are obtained to refine the vertical distribution of cleat geometry
ship, and applicability must be investigated for each sample in ques- in the coal intervals.34 These data are used to improve understand-
tion. ing of the vertical distribution of permeability and the diffusion co-
For the Southern Ute–Mobil 36-1 isotherm, pL = 309.0 psia and efficient in the coal reservoirs because permeability and diffusivity
VL = 733.9 scf/ton. The initial reservoir pressure obtained from should be inversely proportional to cleat spacing.
DST data was 1,366 psia. Based on the isotherm, the pure coal-gas The most useful parameters are the orientation of the fracture sys-
storage capacity is 598.5 scf/ton, which agrees with the pressure- tems that will tend to control flow paths through the reservoir and
core result of 601.4 scf/ton. Comparison of the pressure-core and the natural fracture porosity. The volume of water that will eventual-
isotherm data suggests that the coal is “saturated” in this part of the ly be produced from a coal-gas reservoir is directly proportional to
San Juan basin. Coal is considered saturated when it contains a vol- fracture-system porosity. Porosity can also be estimated by resatu-
ume of gas equal to the storage capacity. rating cores during laboratory flow experiments or by history
The major difficulty encountered in measuring the sorption iso- matching coal-gas well-water production performance with a reser-
therm is the reproduction of the coal moisture content at reservoir voir simulator.
conditions. The coal moisture content is the mass of water sorbed PV compressibility also influences water production. This prop-
onto the coal matrix, not the free water contained in the natural frac- erty has been reported to range from 20 10–6 (see Ref. 35) to
ture system. Sorption isotherm behavior has been shown to be a 8.8 10–4 psi–1 (see Ref. 36). We found that values w10–4 psi–1 are
function of the moisture content during experiments on Appala- required for agreement with estimates derived from interference test
chian basin coals.31 As the moisture content is increased to a critical data.34,37
level, the gas storage capacity is decreased. Our research has shown
that the same phenomenon occurs in Fruitland formation coals in the Relative Permeability. Relative permeability data are required to
San Juan basin.32 When the moisture content exceeds the critical predict the deliverability of gas and water rates as a function of time.
level, the storage capacity is approximately equal to that at the criti- Recent efforts have led to improved estimates of the relative perme-
cal moisture level. We frequently have found that, during experi- ability behavior of the coal natural fracture system for Fruitland
ments, commercial coal-analysis laboratories tend to use moisture formation coal samples.38,39 Current laboratory procedures rely al-
contents that exceed the in-situ values, which results in gas storage most exclusively on unsteady-state techniques as a result of the
capacity estimates that are too low. The ASTM equilibrium-mois- small PV’s of the coal samples and the tendency of the coal to sorb
ture-content determination33 is generally believed to yield the most tracers used to determine saturation distributions during steady-
accurate estimate of the in-situ condition. We recommend this pro- state experiments.
cedure for specification of the isotherm conditions. Measured or published data are most often reported for water sat-
urations from 20% to 80%. Apparent water saturations required for
Fracture Description. We analyzed the coal-reservoir natural frac- well-test analysis or reservoir simulation are frequently outside this
ture (cleat) system by use of a combination of conventional whole- range. We evaluated the applicability of relative permeability rela-
core fracture analysis and polished-coal-block fracture-analysis tions in the literature40,41 to match the published data in the range
methods. Whole-core fractures are analyzed to determine the of the measured saturations and for extrapolation of the relations to
Fig. 1—Graph: comparison of core and log derived ash content, Well Southern Ute–Mobil 36-1.
Fig. 2—Graph: comparison of core and log derived gas content, Well Southern Ute–Mobil 36-1.
the required saturation range. The equations matched the data after Openhole Logging Program and Analysis
slight modification and are as follows. To estimate reservoir properties, openhole logging data are pre-
ƪ ƫ
ferred over core-sample collection and analysis. Interpretation of
2)I
nȀ
l openhole log data is usually less expensive and much faster than use
k rg + k rgǒ1 * S *wǓ 1 * ǒS *wǓ I
l , . . . . . . . . . . . . . . . . . (4) of the other method. A significant number of openhole logs was col-
lected for the wells in this study. This section discusses the logging
program and illustrates methods of estimating ash and gas content
and depths of permeability development.
2)3I
l Table 4 lists the logs that were studied during the Western Creta-
k rw + ǒS *wǓ I
l , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (5) ceous Coal Seam Project and includes the purpose and relative im-
portance of each log. The most useful openhole log in the wells stu-
S w * S iw died was the high-resolution bulk-density log because proper
and S *w + . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (6) interpretation can yield quantitative estimates of ash and gas con-
1 * S iw
tents. The microlog was useful for identifying intervals with perme-
The difference between these relations and those previously pub- ability development in mud-drilled, overbalanced wells. Separation
lished is empirical exponent nȀ. Refs. 40 and 41 presented a value of the shallow, medium, and deep resistivity traces helped to indi-
of nȀ equal to 2. A value of 0.5 more closely matched the published cate permeability development. Other logs, such as the gamma ray,
photoelectric factor (PEF), neutron, and deep resistivity logs, were
data. Generally, we found that Il =100 can be used to approximate
useful for correlation or for recognition of coal seams but were of
measured data because reported data are insensitive to this parame-
lesser value for quantitative analysis.
ter. This value reduces the exponent above the S *w term to [1 (1.02) The logs marked with a dagger ([) (gamma ray, density, and mi-
in Eq. 4 and to [3 (3.02) in Eq. 5. The k gr term is approximately crolog resistivity) in Table 4 are the recommended minimum log-
equal to the relative permeability to gas at irreducible water satura- ging suite for coal evaluation if wellbore conditions permit. Caliper
tion. Estimates of this parameter have ranged from 0.63 to 0.9. Irre- and cable tension traces also should be recorded and are necessary
ducible-water-saturation values range from 0 to 0.3. for quality control of the recorded data. Proper reservoir evaluation
Irreducible-water-saturation estimates are typically low for Fruit- includes analysis of the noncoal lithologies associated with the coal.
land coal natural fracture systems. The laboratory experiments mea- The data marked with a double asterisk (**) in Table 4 must be
sure the pressure loss and flow rate through the part of the fracture known to evaluate other rock lithologies. Because of the thinly
system porosity through which flow occurs during the experiment bedded nature of the Fruitland coal seams, a vertically focused in-
and not in dead-end macropore spaces or micropore spaces in the duction resistivity log in combination with a shallow-focused resis-
coal matrix. The relative permeability to gas at the irreducible water tivity log is preferable to conventional dual-induction-log combina-
saturation is less than the absolute permeability (usually determined tions. Laterologs, which would yield excellent estimates of
when Sw =1.0), even at low irreducible water saturations, because it coal-seam resistivity, are difficult to interpret in water-saturated
is difficult to force gas to flow through the same paths as water. sand intervals because freshwater drilling fluids frequently are used.
The use of more-saline mud systems results in laterolog data that
have been shown to determine intervals of greater permeability ac-
Bulk Density. A CAT scan is used to determine the 3D bulk-density
curately.42
distribution of the core samples. These data are highly useful for de- To determine the utility of wireline-log and core data, the depth-
termination of the ash content of the coal and are commonly used shifted proximate, ultimate, and gas-content core data and the log
to select samples for additional measurements, for core reconstruc- data were investigated statistically. Depth control is a major difficul-
tion, and for core/log depth shifting. Core gamma scans frequently ty encountered when comparing coal-seam-reservoir openhole log
are performed for depth-shifting purposes and reconstruction but do data with core data. Poor depth control is often the result of loss of
not have the vertical resolution of the CAT-scan device. A typical coal samples during coring and while retrieving the core to surface.
CAT-scan resolution is a 0.39-in.-thick vertical slice perpendicular Depth control is further complicated by the cutting of the cores into
to the core axis. 10- or 12-in. sections for the desorption experiments. The cutting
process can account for the loss of 9 in. of core over a 30-ft interval in the coal does not have the same composition as the rocks that sur-
and can cause core-reconstruction difficulties. Depth shifting was round the coal. The ash- and coal-density values are expected to
performed during this study by comparing core-density values ob- vary in other coal basins and must be determined for each reservoir
tained from CAT-scan analysis with the high-resolution openhole of interest. Because of excellent statistical correlations, ash content
density log data. can be estimated from the openhole or CAT-scan bulk-density data
Regression coefficients were computed for each log and core on the basis of the following relation:
relationship for the data collected from three wells. See Table 6 in ò b* ò c
the original preprint of this paper (Ref. 43) for a listing of the regres- f ad + . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (8)
sion coefficients for one well. The most significant correlation òa * òc
achieved from all well data was between the log density and core ash Fig. 1 illustrates the excellent comparison between the log- and
content. Statistical evaluation of the relation between the log- core-derived ash-content estimates on a foot-by-foot basis for Well
derived bulk density and the measured ash content is based on the Southern Ute–Mobil 36-1. Because of the correlations between core
following equation. ash and gas contents and between core ash and openhole bulk-densi-
ty data, it is possible to estimate gas content accurately from the
ò b + ò c ) ( ò a * ò c)f ad. . . . . . . . . . . . . . . . . . . . . . . . . . . (7)
high-resolution bulk-density log data. To account properly for the
Table 5 lists the results of the statistical analyses of Eq. 7 and the ash-free gas-content value of 601.4 scf/ton, which was estimated
log and core data. The estimated coal densities (1.14 to 1.25 g/cm3) from the pressure-core data, in Well Southern Ute–Mobil 36-1, the
agree with those reported in the literature44 for high-vitrinite-con- following relation was used:
tent coal. The Fruitland coal samples are approximately 90% vitri- G c + 601.4 * 751.8f ad. . . . . . . . . . . . . . . . . . . . . . . . . . . (9)
nite. Inspection of CAT-scan data indicates that the lowest-density
coal intervals in the wells have density values between 1.22 and 1.27 The second coefficient, 751.8, was chosen so that the predicted
g/cm3. The estimated ash density is approximately equal to that of gas content was negligible at an ash content of 80%, as suggested
pure montmorillonite clay (2.15 g/cm3) at in-situ conditions.45 by the conventional core-analysis data. Fig. 2 illustrates the excel-
High-ash-content intervals are observed in both the polished-block lent foot-by-foot comparison of the log-derived and core-desorp-
descriptions and CAT-scan bulk-density data. CAT-scan ash densi- tion gas-content estimates. The procedure of determining the rela-
ties generally agree with the statistical estimates. The ash material tion from core data in the area of interest is far superior to relying