Professional Documents
Culture Documents
Johnson 2010
Johnson 2010
Johnson 2010
Enhanced Production Profiling of Gas Wells Using a New Method to Model the
Geothermal Temperature Gradient
M.D. Johnson, SPE, ProTechnics Division of Core Laboratories; M.G. Briceño, Oxy USA Inc.; A.C. Hutchings, SPE,
and D.R. George, SPE, ProTechnics Division of Core Laboratories
This paper was prepared for presentation at the SPE Annual Technical Conference and Exhibition held in Florence, Italy, 19–22 September 2010.
This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not been reviewed
by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers, its officers, or
members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is
restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.
Abstract
Quantitative gas and water production profiling is very useful in completion and production performance evaluation.
Production profile analysis that incorporates mass rate modeling of the flowing temperature has dramatically increased
confidence in profiling results. Accurate mass rate modeling requires an accurate estimate of the geothermal
temperature gradient. The conventional technique of estimating a linear geothermal gradient is not reasonable in
producing regions with long completion intervals and complex lithologies. This paper will detail a new method of
simulating the geothermal temperature gradient and advances in quantitative production profiling.
A multi-well study was conducted in the Piceance Basin in Northwestern Colorado because of the unique effect of
lithology on the geothermal gradient. Coal, sand and shale sequences which impact the geothermal gradient must be
taken into consideration. The study includes open and cased hole log data to define lithology and shut-in temperature
data to characterize the geothermal gradient. An analytical simulator was developed to model the geothermal gradient.
Research in rock, mineral and fluid thermal conductivities is utilized to qualify the field data and analytical model.
Results demonstrate that higher shale volume and coal beds increase the geothermal gradient above clean sand
layer gradients. Measured geothermal gradients across coal bearing intervals are an order of magnitude higher than
clean sand intervals. Geothermal gradients from the simulator were used as constraints in production profile analyses.
The analyses will illustrate the identification and quantification of zonal gas and water production, even when some of
the wells were flowing under adverse conditions of low gas rate, severe slug flow and high water production.
Enhanced production profiling has enabled the operator to enhance their reservoir characterization, completion
strategies and production performance of future offset wells. Other applications for the technology include: 1)
production profiling in similar regions with complex lithology and non-linear geothermal gradient, and 2) distributed
temperature sensing used for continuous temperature and production profile monitoring.
Introduction
Production profiling is the determination of zonal entry rates and non-entry zone flow rates of gas, oil and water at
each depth of the borehole. Accurate quantitative production profiling is critical for completion and production
performance evaluation and reservoir characterization.
The energy industry has used various methods to quantify profile rates. Most methods require the deployment of a
logging tool with a minimum set of tool sensors to measure and record wellbore pressure, temperature, mixture
velocity and phase holdups. Since the measured downhole flow rates require conversion to surface rates, pressure-
volume-temperature (PVT) correlations are used in the analysis. Holdups are commonly established from density
and/or fluid capacitance data. Mixture velocity and total flow rate are estimated from flowmeter data after downhole
calibration. Flow regime zones must be defined. A parameter, slip velocity, is estimated using theoretical and empirical
slip models to calculate the velocity of each phase in a multi-phase flow system. A paper by Ding et al. (1993)
describes this approach but also outlines the concept of an intelligent computer program to enhance slip velocity
determination and improve the accuracy of production profile results.
2 SPE 135321
These strategies all focus on an analytical flow solution using only limited input data and tool data from non-flow
entry zones of the wellbore. The production profile analysis program used in this study is based on a global
optimization, probabilistic well model (Gysen 2008a). The well model can be very complex. Some of the features are
wellbore geometry and hardware, material balance, flow rate constraints, tool response equations, holdups, PVT
inputs or correlations, fluid properties, flow regime models, slip velocity models, reservoir properties and formation
thermal properties. Tool response equations predict the theoretical response of various tools using clearly defined
design and input parameters. These theoretical responses are then compared to the actual measured data and the
well model is modified until the best possible match is achieved. The model is optimized through an iteration process
to minimize the error between the theoretical and measured data for all depths of the wellbore to produce the most
accurate production profile (Donovan et al. 2008).
A temperature module is included and is based on an enthalpy energy conservation model, taking into
consideration: kinetic energy, frictional heating, Joule-Thomson heating or cooling, as well as conduction and
convection heat transfer between the flow stream fluids and formation rock (Gysen 2008b). Temperature responds to
the mass flow rate and thermal heat capacity of the potential flowing phases of gas, oil and water. The differences in
density, especially at low pressures, and heat capacity between gas and water are very significant. This difference
enables quantitative profiling of relatively low water rates. Flowmeter, density and pressure data are also coupled with
the temperature model to quantify a two-phase production profile. Fluid capacitance data are required for three-phase
analysis. The addition of the temperature model improves the accuracy and confidence in production profiling and the
profile results presented are based on this approach.
One of the key parameters required for accurate temperature modeling is the geothermal gradient. The geothermal
gradient determines the temperature of fluids entering the wellbore, when adjusted either cooler or warmer for Joule-
Thomson effects. The flow stream temperature changes during upward or downward movement because of the heat
transfer rate between the flow and near-wellbore, affected in part by the geothermal gradient. In most conventional
reservoirs and tight gas plays, the standard method of estimating a linear geothermal gradient across the gross pay
interval yields reliable results. However, a linear gradient is not reasonable for the tight gas play in the Piceance Basin
in Northwestern Colorado. The basin has a complex lithology sequence with formation layers of differing thermal
conductivities, resulting in a very non-linear geothermal gradient.
This paper will detail a new method of simulating the geothermal gradient to enhance production profiling results.
Production profile comparisons using both linear and simulated geothermal gradients will be presented to illustrate the
value of this method.
chemical flowback tracers. Some wells have formation water breakthrough, which reduces production efficiency and
increases production cost. Because of these factors, the production profile logs are usually conducted at least two
months into production. This delay provides the best opportunity to evaluate the production profile results on a long-
term, production performance basis.
Fig. 1 – Depositional and stratigraphic makeup of the Piceance Basin (Yurewicz et al. 2008).
Initial gas production can reach over 5,000 Mcf/D and steadily decline to under 2,000 Mcf/D when the profile log is
conducted. Weaker wells producing below 1,000 Mcf/D can exhibit severe slug flow conditions downhole. Slug flow
can adversely affect the flowmeter and density data and reduce the accuracy of the profile results. Severe slug flow
conditions coupled with the above mentioned completion and reservoir factors make the Piceance Basin one of the
most challenging areas for quantitative production profiling applications.
Methodology
Because of the complexities affecting thermal conductivity in formation layers, controlled static wellbore
temperature logs were necessary to calibrate the output of a geothermal temperature simulator. Temperature logs
were evaluated to yield correlation between lithology and geothermal gradient. Research on thermal conductivity of
rocks, minerals and fluids was conducted to qualify the observed correlations between lithology and geothermal
4 SPE 135321
gradient. The geothermal temperature simulator was developed with control inputs to allow the best possible match
between the measured static temperatures and simulated geothermal temperatures. Finally, the simulated geothermal
temperature data were used as constraint parameters in the production profile analyses. The results for several wells
are presented in this paper.
Williams
Fork
Cameo
Rollins
Fig. 2: Valley Wells: Gamma ray, shale volume, coal flag and static temperature data, depth correlated to the top of Rollins datum.
SPE 135321 5
Williams
Fork
Cameo
Rollins
Fig. 3: Mesa Wells: Gamma ray, shale volume, coal flag and static temperature data, depth correlated to the top of Rollins datum.
The Cameo is often called the Cameo coal zone because of the multiple layers of coal across this interval. Coal
has very low thermal conductivity when compared to sandstone and shale sedimentary rock. Herrin and Deming
(1996) conducted an extensive study of thermal conductivity of coals. Coal samples from 55 locations across the
United States were tested, including lignite, subbituminous, bituminous and anthracite. The samples obtained from the
Pennsylvania State University Coal Bank were tested as randomly oriented aggregates. Hence, the measured thermal
conductivities are an average of the conductivity both perpendicular and parallel to bedding plane. Herrin and Deming
suggest that the reported numbers should represent the upper limit for coal conductivity perpendicular to bedding,
which determines subsurface temperature gradients. The thermal conductivities range from 0.127 to 0.318 btu/ft/oF/h
with an average mean of 0.191 btu/ft/oF/h. Such values are an order of magnitude lower than most other formation
rock. Geothermal gradient is inversely proportional to thermal conductivity. If the average coal thermal conductivity is
10 times lower than most rocks, then the geothermal temperature gradient would be at least 10 times higher than most
rocks.
None of the coal samples in the study came directly from the Piceance Basin. However, the Cameo coal is
classified as bituminous, of which 36 samples matched the bituminous rank with an average mean thermal conductivity
of 0.174 btu/ft/oF/h. One sample was sourced from the Sunnyside coal, Blackhawk Formation of the Mesaverde Group
from the Uinta Basin in Carbon County, Utah. Johnson and Roberts (2003) describe the various geologic and
structural similarities between the Uinta and Piceance Basins. Sunnyside coal is older than the Cameo coal but should
have common factors other than coal ranking. Therefore, the Cameo coal thermal conductivity may approximate the
Sunnyside coal thermal conductivity at 0.156 btu/ft/oF/h. Thermal conductivity was shown in the coal study to correlate
with the composition of the coal sample. Thermal conductivity is proportional to moisture (water) and ash content and
inversely proportional to the sum of carbon plus volatile content. The Sunnyside coal sample is low in moisture content
at 3.2 %. The Cameo coal layers are considered low in water content and high in carbon (gas) content, which should
exhibit a very low thermal conductivity. A summary of the coal study results are presented in Table 1.
6 SPE 135321
Table 1: Thermal Conductivity of sampled U.S. Bituminous Coals (Herrin and Deming - 1996)
Seam Moisture Content Carbon Content Ash Content Volitile Matter Matrix Density Thermal Conductivity
(not formation name) Location ASTM Rank % Mass % Mass % Mass % Mass g/cc btu/ft/degF/h
Upper Kittanning Armstrong, PA High Volitile A Bituminous 1.3 45.7 23.0 30.0 1.48 0.214
Coal Basin Gunnison, CO Med. Volitile Bituminous 1.5 66.4 5.3 26.8 1.31 0.127
York Canyon Colfax, NM High Volitile A Bituminous 1.6 51.0 11.2 36.2 1.36 0.150
Sewell Greenbrier, WV Med. Volitile Bituminous 1.9 69.5 4.1 24.5 1.31 0.144
Upper Kittanning Barbour, WV High Volitile A Bituminous 1.9 55.8 10.3 32.0 1.35 0.150
Pittsburgh Washington, PA High Volitile A Bituminous 2.5 49.9 9.2 38.4 1.37 0.156
Pittsburgh Greene, PA High Volitile A Bituminous 2.9 52.2 10.0 34.9 1.34 0.156
Elkhorn 3 Floyd, KY High Volitile A Bituminous 3.0 55.3 5.3 36.4 1.32 0.150
Lower Sunnyside Carbon, UT High Volitile A Bituminous 3.2 50.3 9.8 36.7 1.34 0.156
Banner Dickenson, VA High Volitile A Bituminous 4.4 51.6 13.0 31.0 1.40 0.162
Lower Kittanning Somerset, PA Low Volitile Bituminous 4.8 68.5 9.8 16.9 1.40 0.150
Blind Canyon Emery, UT High Volitile A Bituminous 4.9 47.3 5.6 42.2 1.29 0.150
Blind Canyon Emery, UT High Volitile A Bituminous 4.9 43.3 13.5 38.3 1.36 0.179
Pee Wee Morgan, TN High Volitile A Bituminous 5.3 52.3 7.6 34.8 1.37 0.162
Lower Hartshorne Sebastion, AK Low Volitile Bituminous 5.3 65.5 10.6 18.6 1.40 0.156
Splashdam Dickenson, VA Med. Volitile Bituminous 5.4 56.9 11.7 26.0 1.37 0.156
Pratt Walker AL High Volitile B Bituminous 6.1 43.7 19.4 30.8 1.51 0.202
Juanita C Gunnison, CO High Volitile B Bituminous 6.9 50.1 5.2 37.8 1.32 0.156
Mary Lee Walker AL High Volitile B Bituminous 7.2 47.5 19.4 25.9 1.46 0.196
Kentucky 9 Union, KY High Volitile B Bituminous 7.8 42.9 11.3 38.0 1.37 0.179
Pocahontas 3 Buchanan, VA Low Volitile Bituminous 7.8 42.9 11.3 38.0 1.37 0.179
Hiawatha Sevier, UT High Volitile C Bituminous 7.9 47.1 7.9 37.1 1.37 0.179
Ohio 5 Mahoning, OH High Volitile B Bituminous 8.0 52.7 4.4 34.9 1.34 0.144
Ohio 6 Carrol, OH High Volitile B Bituminous 8.2 46.7 9.1 36.0 1.39 0.162
W. Va. 5 Block Kanawha, VA High Volitile B Bituminous 8.4 47.6 14.7 29.3 1.40 0.196
Illionois 6 Perry, Il High Volitile C Bituminous 9.4 44.7 9.5 36.4 1.38 0.185
Tebo Henry, MO High Volitile C Bituminous 9.6 34.5 26.6 29.3 1.57 0.237
Wadge Routt, CO High Volitile C Bituminous 10.9 45.4 6.3 37.4 1.35 0.179
Illionois 6 Randolph, Il High Volitile C Bituminous 11.5 40.4 14.3 33.8 1.41 0.202
Illionois 6 Macoupin, Il High Volitile C Bituminous 11.6 40.5 11.8 36.1 1.37 0.202
Indiana 6 Sullivan, IN High Volitile C Bituminous 12.0 41.3 13.4 33.3 1.41 0.202
New Mexico 8 San Juan, NM High Volitile C Bituminous 12.1 36.4 18.3 33.2 1.46 0.225
Green Navajo, AZ High Volitile C Bituminous 13.0 42.3 5.3 39.4 1.33 0.179
Blue McKinley, NM High Volitile C Bituminous 15.6 40.8 5.3 38.3 1.31 0.179
Brazil Block Greene, IN High Volitile C Bituminous 16.8 47.1 4.4 31.7 1.28 0.179
Adaville 1 Lincoln, WY High Volitile C Bituminous 18.4 40.6 3.4 37.6 1.31 0.179
Averages 7.33 48.80 10.59 33.28 1.37 0.174
dT
qy = ky ⋅ (1)
dy
dT
Γ= (2)
dy
Where q y is the heat flow (btu/ft2/h) in the vertical direction, k y is the thermal conductivity (btu/ft/oF/h) in the vertical
direction of the wellbore and dT is the temperature change (°F) over the distance dy (ft) along the wellbore, or
defined moving forward as the geothermal gradient, Γ .
The heat flow regionally can be assumed constant for the accuracy of this application. Observed gradient changes
in the static temperature data are caused by the different thermal conductivities of each layer of the formation. Bulk
thermal conductivity of a rock is determined by the thermal conductivity of both the rock matrix composition and the
SPE 135321 7
fluid saturations in the pore volume, plus solids from cementation. Most papers report a broad range of thermal
conductivity for various rock types such as sandstone, limestone, dolomite and shale. Clauser and Huenges (1995)
present thermal conductivity ranges for sedimentary rocks such as sandstone and shales. Clauser and Huenges state,
“A table of thermal conductivity purely according to rock type cannot provide site-specific information”. Other stated
factors in determining thermal conductivity are pressure (compaction effects), temperature, degree of saturation and
anisotropy of the rock. A detailed review of the literature on thermal conductivity of rocks confirmed the complexity of
the problem. A list of thermal conductivities for rock types, minerals and fluids from various literature sources are
presented in Table 2 (Clauser and Huenges 1995; Côté and Konrad 2005; Herrin and Deming 1996; Horai 1971;
Midttømme and Roaldset 1999; Prasad et al. 1984).
*Methane thermal conductivity measured at 224 °F and 1,460 psi (approximant average
conditions during production profile log runs in study area).
The major design goals for the geothermal temperature simulator were: 1) compatibility with various types of log
data such as open hole and pulsed-neutron log (PNL) data; 2) user-friendly and efficient operation; and 3) yield
simulated geothermal temperatures accurate enough for production profile analysis applications. Detailed evaluation of
the static temperature data forced the attention on the coal and shale effects. Figure 4 shows the static temperature for
a single example well (code #23) with clear effects from coal layers and shale volume. The depth interval covers the
completed section of the wellbore. Gamma ray and computed shale volume data are presented to indicate the level of
shaliness. Coal layer boundaries, in this case, are identified from the PNL computed data, hydrogen yield capture–
near. The data is normalized for the reservoir parameters of the Piceance Basin and coal layers are flagged for values
greater than 0.18. Below 6,600 ft, the Rollins marine sandstone interval has a temperature gradient of 0.0145 °F/ft.
The temperature gradient increases abruptly to 0.0430 °F/ft across the Cameo zone of the Williams Fork, where the
major coal layers are identified. The coal layers are flagged on Figure 4. Above the Cameo, the lithology transitions
from coal bearing strata to significant shale with interbedded thin sand layers. This interval from 5,050 ft to 6,000 ft
yields a relatively linear temperature gradient of 0.0203 °F/ft. Above 5,050 ft, the lithology changes to thicker, cleaner
sand layers with a major decrease in overall shale volume, resulting in a much lower temperature gradient at 0.0170
°F/ft.
8 SPE 135321
Fig. 4: Well Code #23: Comparison of lithology and static temperature, including average temperature gradients for each lithology type.
Evaluation of open hole log data can provide parameters such as lithology, porosity and saturation to use in
estimating the thermal conductivity of rock layers. The standard reservoir evaluation log conducted in this field is the
PNL. PNL interpretation also provides lithology and synthetic data for porosity and water saturation. Porosity and
saturation parameters assist in estimating the thermal conductivity effects of the fluid in the pore space (Hurter et al.
2007). However, the porosity is usually below 10 percent, which reduces the effects of the pore fluid. Sandstone layers
normally show high gas saturation across the completion zone of the Williams Fork formation. Based on these
observations, the geothermal temperature simulator was not designed to account for the minor changes in porosity
and fluid saturation. Rather, the focus is on the simulation of the effects of rock matrix composition. The geothermal
gradient is simulated based on two conditional relationships for coal layers and non-coal intervals in a basic script
program. Non-coal intervals are defined as:
Where, f sh = fractional shale volume (clean sandstone baseline to 100% shale baseline)
Vsh = shale volume fraction ( Vsh = 1, at 100% shale baseline)
Vsh −clean = shale volume at clean sandstone baseline
The borehole temperature, Tbh , is computed starting at the shallowest reference depth and temperature. The
simulated geothermal gradient from equation 3 or 5 determines the next temperature, defined as
(Γsim )n−1
(Tbh )n = (Tbh )n−1 + (6)
Ls
The simulated geothermal temperature data are computed from Equation 6 for the entire zone of interest. Top and
bottom depth inputs control the depth range in the simulator. Factors for coal and shale in Equation 3 and Equation 5
are relational controls associated with the thermal conductivity differences in the rock matrix. The factor values are not
direct determinations of the thermal conductivity at each depth. They do however reflect the thermal conductivity
relative to clean sandstone used as a baseline. Equation 3 illustrates how the shale factor has more affect on the
computed temperature gradient as the relative shale volume increases above the clean sandstone baseline. Equation
5 handles coal layers in a very straight forward approach. If the lithology is classified as coal, the thermal gradient
increases above the clean sandstone baseline according to the coal factor.
temperatures. Static temperature data were limited below the Rollins marker depth, which may be a factor in the higher
temperature gradient estimate in the Rollins. An abnormally high baseline thermal gradient would yield lower coal and
shale factors, if the thermal conductivities are truly similar across the study area. When the average coal and shale
factors for all the wells in the study were used in the simulator a very good match was observed (Figure 8). The
maximum difference in the static and simulated temperatures is about 2 °F. Therefore, it is believed that the Rollins
thermal gradient was over estimated for this well. Overall, the simulated geothermal temperatures are accurate enough
for production profile analysis applications, meeting another major design goal for the simulator.
Fig. 5: Well Code #23: Static temperature is compared to simulated temperature (coal factor = 12.4, shale factor = 1.75).
Fig. 6: Well Code #27: Static temperature is compared to simulated temperature (coal factor = 11.1, shale factor = 2.08).
SPE 135321 11
Fig. 7: Well Code #25: Static temperature is compared to simulated temperature (coal factor = 9.3, shale factor = 1.40)
Maximum temperature
difference at about 2 °F
Fig. 8: Well Code #25: Static temperature is compared to simulated temperature using average coal and shale factors for all wells in the
study (average coal factor = 12.75, average shale factor = 2.11).
12 SPE 135321
Simulation Results
The multi-well study consists of a total of 35 wells. Table 3 shows the results from 26 wells. Some of the wells were
removed for various reasons such as questionable quality of the static temperature data or lack of data below the top
of Rollins marker. Parameters used in the simulations are listed, such as the Rollins temperature gradient and shale
volume, along with the coal and shale factors. The percentage of coal for the Cameo zone is computed by adding the
total length of all coal layers for the Cameo divided by the length of the Cameo zone. Since the top of Rollins is the
formation structural marker, the depth and temperature at the top of Rollins are listed. Depth at the top of Rollins is
presented in true vertical depth (ft TVD) and is used to sort the results in order of depth. For this study, the wells drilled
on top of the mesa penetrate the deepest structure of the Piceance Basin. Depths range from 9,297 ft TVD to 5,488 ft
TVD, the deepest mesa well to the shallowest well drilled in the valley. Wells are classified by topography as mesa (6
wells) or valley (20 wells) based on the surface drill location. Results are grouped as: all wells, mesa wells and valley
wells, with computed average and standard deviation for the various parameters and factors presented.
Considering all 26 wells, the percentage of coal in the Cameo zone is about 19 percent. There appears to be a
slight decrease in the coal percentage in the middle of the structure down to 14 percent. Highest percentages
approaching 26 percent are observed in the upper structure or shallower depths. The top of Rollins temperature
ranges from 195 °F to 242 °F, because of the change in depth. Rollins temperature gradient is relatively consistent
with an average of 0.0135 °F/ft for all wells. Mesa and valley wells vary from 0.0112 °F/ft to 0.0141 °F/ft, respectively.
Rollins shale volume is consistent with an average of 0.082 or 8.2 percent for all wells with little change with
topography groups. Coal and shale factors average 12.75 and 2.11, respectively, for all wells. Both factors are higher
for mesa wells at 14.28 and 2.41. The increase is likely caused in part by the lower Rollins temperature gradient used
in the simulations. Coal and shale factors clearly illustrate the relationship between lithology and thermal conductivity.
Their relative values match reasonably close to the published thermal conductivity values for rock type in Table 2. A
thermal conductivity study was conducted in the Paraná Basin in southern Brazil from a large number of core samples
(Hurter and Pollack 1996). Paraná Basin results show thermal conductivity ratios between coal to sandstone and shale
to sandstone at 11.9 and 2.04, respectively. These ratios are very close to the relative ratios for the coal and shale
factors as related to the Rollins Sandstone in the Piceance Basin study at 12.75 and 2.11, respectively.
and 208 BWPD. Figure 10 shows the gas and water production profile analysis results. QGas and QWater represent
the flow rates at standard surface conditions for each depth. Percent contributions for gas and water are listed for each
stage interval. Analysis data for velocity, density, pressure, temperature and simulated geothermal temperature
(labeled as geothermal gradient, DegF) are presented. Radioactive tracer imaging is used for completion diagnostics
on most wells. Figure 10 shows the tracer imaging for antimony-124, iridium-192 and scandium-46 isotopes on tracks
to the left and right of the profile data. Tracers show good frac proppant placement and coverage for all stages. By
coupling the tracer evaluation with the production profile results, the profile can be correlated to the quality of the
formation rock. Tracers can identify a completion efficiency problem such as non-stimulated zones. Without tracers,
there is no way to know with any certainty that all the perforation intervals were stimulated. Any non-traced, post-frac
production profile comes with the possible wrong assumption that all the zones were stimulated and the profile
represents the reservoir characteristics. The second log track from the right show flags for each frac stage interval.
The color code corresponds to the tracer pumped for each stage. The coal layers are identified by the small black flags
just to the right of the stage flags.
Major gas production contribution is detected from Stages 1, 2 and 3 (6,056 to 6,580 ft) at 76% of the total gas.
Velocity increases and density decreases across these stages identify major gas production. The temperature cooling
relative to the geothermal gradient is caused by minor Joule-Thomson cooling effects and flow mixture response.
Joule-Thomson effect is usually observed in gas producing wells as cooling because of gas expansion from a pressure
change (Steffensen and Smith 1973). Flow entry at any depth should enter the wellbore at the geothermal
temperature. Gas entry usually exhibits some level of cooling below the geothermal gradient temperature caused by
Joule-Thomson effect. High conductivity fractures should not exhibit major cooling unless near-wellbore damage is
creating a pressure drop. The temperature model match does not indicate any significant damage. The observed
cooling across Stages 2 and 3 is caused by significant gas entries at or below the geothermal gradient and resulting
flow stream mixture temperatures. Stages 1 and 2 are completed in the Cameo and Stage 3 is in the transitional zone
of the lower Williams Fork with significant coal layering. Mass rate temperature modeling using the new geothermal
gradient enhances the accuracy of the gas profile, especially when the velocity data are adversely affected by slug
flow.
At 100% flow above the top perforations, the volumetric downhole flow rates are about 88% gas and 12% water.
However, the mass downhole flow rates are about 28% gas and 72% water. Velocity and density strongly respond to
the dominant volumetric rate of the flow stream, which is the gas flow in this well. Temperature, on the other hand,
responds to the dominant mass rate or the water flow. Major water production contribution is detected from the upper
Williams Fork from Stages 8, 9 and 10 (4,522 to 5,037 ft) at 86% of the total water. Temperature cooling across these
zones is resolved in the mass rate temperature model as major water entry zones with trace gas production. The slope
of the temperature data decreases significantly above the water entry zones because of the increase in mass rate and
heat capacity of the flow stream. Accurate quantitative evaluation would not be possible without a reliable simulated
geothermal gradient. Water enters the wellbore at the geothermal temperature, assuming negligible Joule-Thomson
heating effect. The temperature model can not properly balance the energy conservation for the system without an
accurate geothermal gradient. The density increases across Stage 8 indicate significant water entry. However, the
density does not respond to the water entries from Stage 9 and 10. The velocity data are certainly too noisy to identify
these water entry zones which only make up a small percentage of the volumetric flow stream. The ability to identify
and accurately quantify the water profile using temperature modeling illustrates the value of this new method.
Stage 10
Stage 1
Fig. 10: Well Code #23: Production profile results with analysis data.
Figure 11 combines the profile results with the reservoir evaluation data from the PNL log for the Code #23 well.
The reservoir evaluation data are defined as: CORE_PPH is synthetic porosity, CORE_SUL is water saturation,
CH_BVWN is the product of CORE_PPH and CORE_SUL, and RESFLAG identifies the target reservoir pay or
completion zone based on defined cutoff values. Sand layers become thicker from Stage 8 and above as detected
primarily from the porosity. At the time of the completion, these zones appeared to be good quality sands with good
potential for gas production. Unfortunately, on the basis of the production profile results on this and other wells in the
field, high water production is a major risk when this interval of the Williams Fork is completed. Based on the geologic
understanding of this interval, the sands are much more continuous laterally in contrast to the lenticular structure of the
16 SPE 135321
Williams Fork below. The connectivity of the upper Williams Fork sands can allow gas migration out and water
migration into sands. Evidence of gas migration has been observed from lower pore pressure measurements during
pre-frac injection testing. The estimated top of continuous gas is about 5,325 ft MD, which is below these water
producing zones. Hood and Yurewicz (2008) emphasize the high risk of water production from sands above the top of
continuous gas in the upper Williams Fork Formation.
Stage 10
Stage 1
Fig. 11: Well Code #23: Production profile results with reservoir evaluation data.
The initial production profile analysis on this well used a linear geothermal gradient at about 0.024 °F/ft. Figure 12
illustrates the comparison of the gas and water profile results of Model 1, using a linear geothermal gradient, versus
Model 2, using the simulated geothermal gradient as presented in Figure 11. The water profile changed dramatically
between Model 1 and Model 2 with more water production contribution across Stages 7, 8 and 9 for Model 2. Model 1
placed more water from Stages 3, 4 and 5 because of the poor geothermal gradient estimate. The overall gas profile
on a frac stage basis did not change much between Model 1 and Model 2. However, the Model 2 results do show
better gas profile resolution. Model 1 required more near-wellbore pressure drop to match the amount of Joule-
Thomson cooling because the flowing temperature is below the linear geothermal gradient from Stage 1 to Stage 6.
Increased cooling and pressure drop is an indication of some level of skin damage. The Model 2 temperature
evaluation required only a minimal pressure drop with no skin damage effects evident.
SPE 135321 17
Temperature
Compensated
geothermal gradient
Major change in water
profile with compensated
geothermal gradient
Top of
Rollins Linear geothermal gradient
Fig. 12: Well Code #23: Production profile comparison between Model 1 (linear geothermal gradient) and Model 2 (simulated geothermal
gradient).
Stage 10
Stage 1
Fig. 14: Well Code #27: Production profile results with analysis data.
SPE 135321 19
Stage 10
Stage 1
Fig. 15: Well Code #27: Production profile results with reservoir evaluation data.
20 SPE 135321
Stage 8
Stage 1
Fig. 17: Well Code #25: Production profile results with reservoir evaluation data.
Conclusions
The geothermal temperature simulator was successful in modeling the static temperature data for the majority of
the wells in the study. Simulation and static temperature comparison results for several wells are presented in Figures
5–8. The simulation results for all the wells in the study are listed in Table 3. The simulator control factors for coal and
shale volume relate closely to the thermal conductivity values for sandstone, shale and coal found in the literature.
Accurate simulated geothermal temperature data greatly enhanced the production profile results by improving the
mass rate temperature modeling process. Production profile results for some of the worst case wells with low gas rate,
high water production and severe slug flow conditions are presented in Figures 9–17.
On the basis of the high confidence in production profile results combined with other reservoir evaluation data, the
operator has enhanced their characterization of the reservoir. Some of the benefits are: 1) identification of primary gas
producing layers, 2) identification of water producing layers, and 3) improved reservoir modeling and volumetric gas-in-
place estimates. Wells drilled in the future may not allow the opportunity to obtain static temperature data prior to
completion operations. This method will allow accurate geothermal gradient determination by using field average
parameters in the simulator and fine-tuning these parameters based on the production profile data and analysis match
to the well model. Coal layer identification and shale volume estimates will be required.
Completion and production optimization has advanced by coupling radioactive frac tracers and production profile
results. Completion strategies are optimized by targeting the main gas sands and avoiding sands with a high risk of
22 SPE 135321
water production. As a result, completion cost has reduced by fracture treating a lower number of stages and zones.
Water handling cost has decreased. The major benefit however, is the improved production performance of individual
wells and the field. Measurement based solutions were applied to this tight gas development area with marked
improvements.
Enhanced production profiling through accurate geothermal temperature simulation can be applied to other
producing regions. Any producing region similar to the Piceance Basin with complex lithology and non-linear
geothermal gradient could use this method. Regional factors may need to be accounted for in the simulator design.
Distributed temperature sensing (DTS) enables continuous wellbore temperature measurement with the deployment of
fiber-optic cable into the wellbore (Johnson et al. 2006). However, any flow profiling analysis program must rely on an
accurate geothermal gradient. Therefore, the geothermal temperature simulator developed for this study could be used
for DTS applications to improve the accuracy of the production profile.
Nomenclature
Antimony-124 = antimony-124 tracer, API
BVG = bulk volume, gas
BVW = bulk volume, water
BWPD = barrel(s) water per day
BFPD = barrel(s) fluid per day
CCL = casing collar locator, mV
CH_BVWN = bound water
Coal = coal layer
Coal Flag = coal flag
CORE_PPH = synthetic porosity
CORE_SUL = water saturation
CPS = counts per second
dy = vertical distance along the wellbore, ft
dT = temperature change, °F
DEGF = degrees Fahrenheit
Density = wellbore fluid density, g/cc (common units for production profile analysis)
ƒsh = fractional shale volume, clean sandstone baseline to 100% shale baseline
ft MD = measured depth, ft
ft TVD = true vertical depth, ft
F coal = coal factor, relational to coal thermal conductivity
F sh = shale factor, relational to shale thermal conductivity
FPM = feet per minute
Geothermal Gradient = simulated geothermal temperature (profile log images), DegF, °F
GR = gamma ray, API, CPS
h = time, hour
Iridium-192 = iridium-192 tracer, API
k = thermal conductivity, btu/ft/°F/h
ky = thermal conductivity in the vertical direction, btu/ft/°F/h
Ls = depth spacing, data samples per foot
Line Speed = line speed, FPM
mV = millivolt
Mcf/D = thousand cubic feet per day
PERF = perforation zone
Pressure = wellbore pressure, psi
q = heat flow, btu/ft2/h
qy = heat flow in the vertical direction, btu/ft2/h
QGas = gas flow rate, Mcf/D
QWater = water flow rate, BFPD
RESFLAG4 = reservoir pay flag
RPS = revolutions per second
Scandium-46 = scandium-46 tracer, API
SPE 135321 23
Acknowledgements
The authors wish to thank the management of Oxy USA Inc. and Core Laboratories for permission to publish this
manuscript. They also express special appreciation to the Piceance Basin Asset Team. Specifically, we would like to
thank Alex McCoy for his vision, guidance and support during all phases of this project.
References
1. Bullard, E. C. 1947. The Time Necessary for a Bore Hole to Attain Temperature Equilibrium. Geophysical Journal International
Volume# 5, (Issue# s5): 127-130.
2. Clauser, C and Huenges, E. 1995. Thermal Conductivity of Rock and Minerals. Rock Physics and Phase Relations: a Handbook of
Physical Constants, Ahrens, T. J. ed., American Geophysical Union, Washington: 105-126.
3. Côté, Jean and Konrad, Jean-Marie 2005. Thermal Conductivity of Base-Course Materials. Canadian Geotechnical Journal 42: 61-78.
4. Ding, Z. X., Flecker, M. J., and Anderson, C. 1993. Improved Multiphase Flow Analysis Using an Expert System for Slip Velocity
Determination. Presented at the SPWLA 34th Annual Logging Symposium 13-16 June 11pp.
5. Donovan, G., Dria, D., Ugueto, G., Gysen, A. 2008. Improved Production Profiling Using Thermal Balance and Statistical Modeling in
the Pinedale Anticline of the US Rocky Mountains. SPE 115034 prepared for presentation at the 2008 SPE Annual Technical
Conference and Exhibition held in Denver, Colorado, U.S.A. 21-24 September: 10 pp.
6. Gysen, M. 2008. Introduction to the Optimization Process. In Production Logging Analysis Tool: PLATO Reference Guide, Edition 4,
Chap. 6.5.1, 148-149.
7. Gysen, M. 2008. Analyzing Temperature Data. In Production Logging Analysis Tool: PLATO Reference Guide, Edition 4, Chap.
6.5.11, 159-162.
8. Herrin, J. and Deming, D. 1996. Thermal Conductivity of U.S. Coals. Journal of Geophysical Research Volume# 101, (Issue# B11):
25,381-25,386.
9. Hood, K. C. and Yurewicz, D. A. 2008. Assessing the Mesaverde Basin-centered Gas Play, Piceance Basin, Colorado. 2005 Vail
Hedberg Conference: AAPG Hedberg Series (Issue# 3): 87-104.
10. Horai, Ki-iti. 1971. Thermal Conductivity of Rock-Forming Minerals. Journal of Geophysical Research Volume# 76 (Issue# 5): 1,278-
1,308.
11. Hurter, S. and Pollack, H. 1996. Terrestrial Heat Flow in the Paraná Basin, Southern Brazil. Journal of Geophysical Research
Volume# 101, (Issue# B4): 8,659-8,671.
12. Hurter, S., Garnett, A., Bielinski, A., Kopp, A. 2007. Thermal Signature of Free-Phase CO2 in Porous Rocks: Detectability of CO2 by
Temperature Logging. SPE 109007 prepared for presentation at the SPE Offshore Europe 2007 held in Aberdeen, Scotland, U.K. 4-7
September: 12 pp.
24 SPE 135321
13. Johnson, D., Sierra, J., Kaura, J., and Gualtieri, D. 2006. Successful Flow Profiling of Gas Wells Using Distributed Temperature
Sensing Data. SPE 103097 prepared for presentation at the 2006 SPE Annual Technical Conference and Exhibition held in San
Antonio, Texas, U.S.A. 24-27 September: 16 pp.
14. Johnson, R.C., Roberts, S.B. 2003. The Mesaverde Total Petroleum System, Uinta-Piceance Province, Utah and Colorado. In
Petroleum Systems and Geologic Assessment of Oil and Gas in the Uinta-Piceance Province, Utah and Colorado, Chap. 7, 1-63.
Denver, Colorado: U.S. Geological Survey Digital Series DDS-69-B
15. Midttømme, K. and Roaldset, E. 1999. Thermal Conductivity of Sedimentary Rocks: Uncertainties in Measurement and Modeling.
Geological Society, London, Special Publications Volume# 158: 45-60.
16. Prasad, R.C., Mani, N., and Verart, J. E. S. 1984. Thermal Conductivity of Methane. International Journal of Thermophysics Volume#
5, (Issue# 3): 265-279.
17. Scheevel, J.R. 2006. Elastic Fracture Analysis in the Tight Gas Sands of the Mesaverde Formation, Piceance Basin, Colorado. AAPG
Annual Convention, Houston, TX, 9-12 April 2006.
18. Steffensen, R. J., Smith, R. C. 1973. The Importance of Joule-Thomson Heating (or Cooling) in Temperature Log Interpretation. SPE
4636-MS prepared for presentation 48th Annual Fall Meeting of the Society of Petroleum Engineers of AIME, held in Las Vegas,
Nevada, U.S.A. 30 September - 3 October: 15 pp.
19. Yurewicz, D. A., Bohacs, K. M., Kendall, J., Klimentidis, R. E., Kronmueller, K., Meurer, M. E., Ryan, T. C., and Yeakel, J. D. 2008.
Controls on Gas and Water Distribution, Mesaverde Basin-centered Gas Play, Piceance Basin, Colorado. 2005 Vail Hedberg
Conference: AAPG Hedberg Series (Issue# 3): 105-136.
20. Zhang, E., Hill, R., Katz, B., and Yongchun, T. 2008. Modeling of Gas Generation from the Cameo Coal Zone in the Piceance Basin,
Colorado. AAPG Bulletin Volume# 92, (Issue# 8): 1,077-1,106.