Академический Документы
Профессиональный Документы
Культура Документы
Summary
Petrographic work on thin sections from rock samples collected
in tight gas sandstones of the western Canada sedimentary basin
(WCSB) shows that the sandstones are composed of intergranular,
microfracture and slot, and isolated noneffective porosities. The
petrographic observations of these triple-porosity rocks have led
to a petrophysical interpretation with the use of a triple-porosity
model.
Tight gas reservoirs are very complex heterogeneous systems
that have been evaluated in the past mostly with single-porosity
models. We propose that these types of reservoirs can be represented better by triple-porosity models for more rigorous quantitative petrophysical characterization. The triple-porosity model
discussed in this paper fits the petrographic observations very
well, leading to a more rigorous characterization of effective and
noneffective porosity.
The petrographic and core-calibrated triple-porosity model is
then used for well-log interpretation of those wells when these data
are not available. The result is a reasonable quantitative characterization of the tight gas reservoir that can be used for improving
hydraulic-fracturing design, flow-units determination, reservoir
engineering, and simulation studies. The data can be determined
at room conditions and simulated conditions of net stress.
It is concluded that honoring with a triple-porosity model the
different types of porosities observed in thin sections and cores
leads to more-rigorous and -useful petrophysical interpretations
that can be linked to gas productivity.
Introduction
The GFREE gas research team at the University of Calgary has
estimated natural-gas endowment in tight gas sands in Canada at
105 Tcf (Aguilera 2010). GFREE stands for an integrated multidisciplinary team researching geoscience (G); formation evaluation
(F); reservoir drilling, completion, and stimulation (R); reservoir
engineering (RE); and economics and externalities (EE). The estimated endowment corresponds to only 7% of the original gas in
place (OGIP) of 1500 Tcf estimated by Masters (1984) and investigated in a sensitivity analysis carried out by Contreras and Aguilera
(2011). Fig. 1 shows the location of the study area. Fig. 2 shows
the stratigraphic column of the study area and corresponding stratigraphy from its northwest (Peace and Pine Rivers) and southeast
extents (southwestern Alberta and southeastern British Columbia).
The Upper Jurassic to Lower Cretaceous Nikanassin group is
generally characterized as tight gas formations with low values
of permeability (typically a fraction of a millidarcy) and low
porosities (usually less than 6%). It is likely that natural microfractures and slot pores dominate the productivity of the formation.
These secondary pores are stress sensitive. However, in those cases
in which fractures and slots are partially mineralized (for example,
with partial quartz overgrowths) the secondary minerals might act
as natural propping agents that help fractures and slots to remain
open. Lack of any secondary minerals might result in partial or
total closure, depending on the in-situ stresses. It is also likely that
in some cases the values of OGIP might be overestimated because
566
petrographic studies indicate that the there are rocks in which part
of the total porosity is nonconnected (isolated and noneffective).
Triple-Porosity Rocks
Fig. 3 shows an example of a triple-porosity rock in a thin section of the Cadomin formation in the WCSB. The section clearly
shows the presence of intergranular porosity, slot porosity, and
micro-fractures cross cutting clasts and sandy matrix*. In this
case, it appears that all porosities might be connected. Previous
petrophysical studies by Solano (2010) and Solano et al. (2010)
indicate that the Nikanassin group is made up of different porosity
types that we classify as (1) intergranular, (2) microfractures and/or
slots, and (3) isolated (noneffective porosity). Fig. 4 shows image
processing of thin-section microphotographs used to analyze the
geometry of individual slots within a tight rock in the WCSB. Figs.
4a and 4b correspond to the original image of a thin section under
transmitted plane-polarized and cross-polarized light, respectively.
Blue staining was used to highlight the porosity of the rock. This
sample corresponds to a fine- to medium-grained, well-sorted
sublitharenite sandstone that contains between 5 and 25% detrital
rock fragments. The abundance of quartz grains and the extensive
quartz overgrowth around them are noticeable. There are also visible isolated partially altered chert and shale fragments, authigenic
kaolinite (probably former feldspar fragments), and detrital mica
fragments forming secondary-microporosity spots. Porosity and
maximum permeability from routine core analysis of an adjacent
sample were reported as 5.1% and 0.18 md, respectively. The
image in Fig. 4c represents results of several filters applied to the
original image to highlight the porosity of the sample. The image
in Fig. 4d is a simplified representation of the pore geometry
observed in the previous images. Vertical and horizontal axes are in
micrometers, and the vertical axis coincides with the stratigraphic
up/down orientation of the sample. These data have been used for
simulations at the pore-scale level (Rahmanian et al. 2010).
Several characteristics of the sandstones cause the effectiveporosity values to be lower than the total porosity including
pyrobitumen in pores; compaction and squeezing of shale clasts
into pore spaces; compaction of quartz and other grains into pore
spaces; common microporosity in chert grains; and shale clasts,
degraded volcanic rock fragments, and quartz overgrowths. The
percentage of micropores vs. mesopores and macropores is commonly greater than 50%, often up to 100%. Although microporosity is generally noneffective porosity, in the case of the rocks
considered in this study, microporosity can be effective if present
in cherts, particularly when examined at the scale of drill cuttings.
At the scale of core plugs it can be partially disconnected. At the
scale of the reservoir it can be either partially disconnected or
completely disconnected. As a result, from the point of view of the
petrophysical triple-porosity model, part of the microporosity can
be effective and part ineffective. The same reasoning applies to the
case of dissolution porosity in which portions of this porosity can
be completely isolated and portions can be communicated through
slots and microfractures.
The relationship between total porosity from core/log calibration and effective porosity is nonlinear and is controlled by several
independent variables, as well as the relation between total porosity
and permeability. Therefore, correction of the log interpretation
with a triple-porosity model becomes critical.
*Personal communication with T. Moslow, 2011, Calgary.
(1 nc )2
log nc +
2 + (1 2 nc ) / b mb
m=
, . . . . . . . . . . . . (1)
log
where is total porosity, fraction; b is matrix block porosity
scaled relative to the bulk volume of the matrix system, fraction;
nc is nonconnected porosity (PHInc) scaled relative to the bulk
volume of the composite system, fraction; and 2 is porosity of
natural fractures (PHI2) scaled relative to the bulk volume of the
composite system, fraction. The development of Eq. 1 has been
presented by Al-Ghamdi et al. (2010) where it was used for evaluation of fractured and vuggy carbonates in the Middle East. The
use of the equation was extended to evaluation of tight gas formations by Deng (2010) and Leguizamon and Aguilera (2011). The
porosities in Eq. 1 can be integrated as follows:
2
; v = nc ;m = 2 nc ;
nc
, . . . . . . . . . . . . . . . . . (2)
b = m / (1 2 nc )
v =
UPPER JURASSIC
Berriasian
?
?
Hauterivian
Valanginian
Gladstone
Southwestern Alberta
and Southeastern
British Columbia
Gladstone
?
?
?
?
?
KootenayGroup
Barremian
Central Alberta
Foothills
LuscarGroup
LOWER CRETACEOUS
Albian
STAGE
Minnes Group
SERIES
BlaimoreGp.
Monteith
Tithonian /
Volgian
Mist Mountain
Morrisey
Kimmeridgian
Oxfordian
Fig. 2Stratigraphic column of the study area and corresponding stratigraphy from its northwest (Peace and Pine Rivers) and
southeast extents (southwestern Alberta and southeastern British Columbia). Adapted from Stott (1998).
October 2011 SPE Reservoir Evaluation & Engineering
567
3 slot
Intg.
Fig. 3Thin section showing intergranular porosity, slot porosity, and microfractures cross cutting clasts and sandy matrix,
Cadomin formationElmworth, Deep Basin*.
568
Fig. 4Image processing of thin-section microphotographs used to analyze the geometry of individual slots within a tight rock
[source: Rahmanian et al. (2010)].
PHI_TS (%):
3
36
69
>9
(a)
PHI_DCs (%):
3
36
69
>9
(b)
Fig. 5Ternary-plot diagrams representing percentages of the principal pore geometries observed on thin sections prepared from
drill-cutting samples. Ranges of porosity values estimated from thin sections and from measurements on drill cutting samples
are used as additional constraint in Figs. 5a and 5b, respectively. Fig. 5a shows a ternary diagram with porosity values estimated
from thin sections (PHI_TS). Fig. 5b shows porosity from laboratory measurements on drill cuttings (PHI_DCs). In this example,
the dissolution and microporosity pores are dominant. Source: Solano (2010).
Dengs simulation (Deng et al. 2011), we use a modification proposed by Aguilera (2010):
rp35 = 2.665[ k / (100 ) ] , . . . . . . . . . . . . . . . . . . . . . . . . . . (3)
0.45
Fig. 6Petrographic data are represented by red triangles. The triple-porosity model permitted separating the data and quantifying porosities and m for three different formations (Petrography courtesy of ConocoPhillips).
flow units. These graphs use, for example, helium porosities and
air permeabilities. These unstressed properties have been compared
empirically with actual oil- and gas-production rates to generate a
wide range of anticipated production outcomes. For example, for
the case of oil wells, Martin et al. (1997) indicate that comparatively megaports can reach medium-gravity-oil production rates
of tens of thousands of barrels per day if zonal thickness and
other factors are constant; and without mechanical constraints,
macroports can reach thousands of barrels per day, and mesoports
hundreds of barrels per day. Microports can produce several tens
of barrels per day on pump. However, Martin et al. (1997) state
that microport flow units are decidedly nonreservoir in this comparative completion of moderate thickness and medium gravity oil
Fig. 7Petrographic data of Formation D are represented by blue squares. A regression is provided by the dashed line. Results
from the triple-porosity model are shown by the continuous red line. Upper and lower bounds of the data are given by the dotted lines (Petrography courtesy of ConocoPhillips).
October 2011 SPE Reservoir Evaluation & Engineering
571
2.20
2.00
1.80
1.60
Form. A
Form. B
1.40
Form. C
Form. D
Core Mesaverde
1.20
1.00
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
Total Porosity
Fig. 8Values of the cementation exponent m for Formations
A, B, C, and D calculated using the petrophysical triple-porosity
model. Core data from the Mesaverde formation in the USA are
taken from Byrnes et al. (2008a, 2008b).
Stress-Dependent Rocks
All rocks are stress dependent to a major or minor degree. For
situations or conditions where there is a need for showing stressdependent porosities, permeabilities, and pore-throat aperture,
laboratory work is the optimum way for evaluating this dependency. Under favorable circumstances, the in-situ permeability
can be estimated from coupled geomechanical reservoir simulation including numerical modeling of flow and buildup tests of
individual wells. The optimum way of accomplishing this task
is by cutting slices of the fully characterized heterogeneous 3D
simulation model and matching the flow and buildup pressures.
This can lead to a significant reduction of permeability in tight-gas
formations with a negligible amount of secondary mineralization.
However, in those cases where there is partial secondary mineralization (for example, partial quartz overgrowths inside slot
porosity), the reduction in the value of in-situ permeability might
be very small (Deng 2010).
The rp35 graph can also be developed for different net-stress
conditions, ideally with the support of core data. The reduction
in porosity and permeability can be calculated with sophisticated
models or with simplified but solid equations. One example is
Fig. 9Log interpretation of well in tight gas formation of the WCSB showing effective porosity (intergranular plus microfractures
and slots) and variable values of m calculated from the petrophysical triple-porosity model.
572
kg/m2
kg/m2
kg/m2
kg/m2
Fig. 10Proppant-concentration (kg/m2) evolution of the actual job in example well showing fracture growth after 25, 35, 50, and
68.5 minutes. Maximum fracture length at the end of the job is approximately 50 m. There is no vertical containment between
the fracture intervals.
shown in Fig. 13, for the case in which the rp35 curve is equal to
0.2 m at room conditions. For this example, changes of porosity
and permeability are calculated from (Jones 1975; Walsh 1981)
k
=
1 k1
1/ 3
log pk log ph .
log pk1 log ph
. . . . . . . . . . . . . . . . . . . . . . (4)
theoretical developments. Both lead to Eq. 4, making it very valuable from theoretical and practical points of view. This equation
takes into account the presence of natural fractures. We use it on
tight gas formations because of the presence of microfractures and
slot porosity. The net stress on the rocks is represented by pk. The
pressure at which a fracture would heal is ph. Subscript 1 is used
to represent the initial net-stress condition. However, any other
equation can be used to estimate the reductions on these properties as a function of net stress. Results are shown in Fig. 13 with
curves developed at room conditions and net stresses equal to 500
573
kg/m2
kg/m2
kg/m2
kg/m2
Fig. 11Proppant-concentration evolution of simulated optimized job in the same well as in Fig. 10, pumping initially 1 t of 100mesh sand. Results are shown at 25, 35, 55, and 68.5 minutes. The fracture length with good proppant concentration at the end
of the job is approximately 80 m. The total length with smaller amounts of proppant going to the bottom of the fracture reaches
more than 160 m. Note that in this optimized job the fracture is contained vertically.
and 5,000 psi. Note from Eq. 4 and Fig. 14 that the reductions
in permeabilities are larger than the reductions in porosity. The
reductions in pore-throat apertures (rp35) are shown in the righthand side of Fig. 13.
Conclusions
1. A petrophysical triple-porosity model has been shown to be a useful
tool for characterization of tight gas formations in the WCSB.
2. The model distinguishes between effective and noneffective
porosity. The effective porosity is the summation of intergranular
574
porosity plus microfractures and slot porosities. The noneffective porosity is nonconnected porosity.
3. The model provides useful information for 3D simulation of
hydraulic-fracturing jobs in tight gas formations. This in turn
permits optimizing hydraulic-fracturing jobs.
4. A preliminar empirical link has been established between rp35
and oil and gas rates.
5. Porosities, permeabilities, and pore-throat apertures calculated
at room conditions can be adjusted to reflect the effects of net
stress in Winland-type crossplots, as needed.
October 2011 SPE Reservoir Evaluation & Engineering
MMscf/D
Thousand
BOPD
rp35
rp35 microns
0.1
10s
Room
100s
1s
PERMEABILITY (md)
0.2
0.1s
10s
0.01s
1s
0.141
500
0.098
0.1s
5000
0.01
0.001
0
Nomenclature
k = permeability, md
k1 = initial permeability, md
m = porosity exponent (cementation factor) of triple-porosity
reservoir
mb = porosity exponent (cementation factor) of only the matrix block
ph = apparent healing pressure, psi (kPa)
pk = net stress, psi (kPa)
pk1 = initial net stress, psi (kPa)
rp35 = pore-throat apertures at 35% cumulative pore volume (35%
mercury saturation during a capillary pressure test)
v = partitioning coefficient, fraction
10
15
20
25
30
POROSITY (%)
Source: GFREE Research Team, U of Calgary, 2011
vnc
Vdis
Vlam
Vstruc
VSLT
1
2
Percent Decrease
1
0.9
Porosity reduction
0.8
Permeability reduction
0.7
rp35 reduction
0.6
0.5
0.4
0.3
0.2
0.1
0
0
1,000
2,000
3,000
4,000
5,000
575
577