Escolar Documentos
Profissional Documentos
Cultura Documentos
ABSTRACT
Fault seal can arise from reservoir/nonreservoir
juxtaposition or by development of fault rock having high entry pressure. The methodology for evaluating these possibilities uses detailed seismic mapping and well analysis.
A first-order seal analysis involves identifying reservoir juxtaposition areas over the fault surface by
using the mapped horizons and a refined reservoir
stratigraphy defined by isochores at the fault surface.
The second-order phase of the analysis assesses
whether the sand/sand contacts are likely to support a
pressure difference. We define two types of lithologydependent attributes: gouge ratio and smear factor.
Gouge ratio is an estimate of the proportion of finegrained material entrained into the fault gouge from
the wall rocks. Smear factor methods (including clay
smear potential and shale smear factor) estimate the
profile thickness of a shale drawn along the fault
zone during faulting. All of these parameters vary
over the fault surface, implying that faults cannot
simply be designated sealing or nonsealing.
An important step in using these parameters is to
calibrate them in areas where across-fault pressure
differences are explicitly known from wells on both
sides of a fault. Our calibration for a number of data
sets shows remarkably consistent results, despite
their diverse settings (e.g., Brent province, Niger
Delta, Columbus basin). For example, a shale gouge
ratio of about 20% (volume of shale in the slipped
interval) is a typical threshold between minimal
across-fault pressure difference and significant seal.
INTRODUCTION
Faults play an important role in creating hydrocarbon traps. For a better appreciation of the risks assoCopyright 1997. The American Association of Petroleum Geologists. All
rights reserved.
1Manuscript received March 7, 1996; revised manuscript received August
12, 1996; final acceptance November 21, 1996.
2 Badley Earth Sciences Ltd., North Beck House, North Beck Lane,
Hundleby, Spilsby, Lincolnshire PE23 5NB, United Kingdom.
We are grateful to Torbjrn Fristad, Dave Phelps, and Jon Arne verland
for detailed discussions of some of the data sets described in this article, and
to Robert Berg and Grant Skerlec for constructive reviews of the initial
manuscript.
ciated with fault-controlled prospects and of the production from faulted fields, it is important to understand the processes that contribute to fault seals.
Given certain information about a fault cutting a
reservoir sequence, it is desirable to predict the likely sealing behavior of each part of the fault system.
Seals can be considered as membrane seals or
hydraulic seals, depending on their likely failure
mode (Watts, 1987). The dominant control on failure of membrane seals is the capillary entry pressure of the seal rock; that is, the pressure required
for hydrocarbons to enter the largest interconnected pore throat of the seal. When the entry pressure
has to exceed the strength of the rock in order to
breach the seal, the seal is considered a hydraulic
seal. A number of mechanisms have been recognized whereby fault planes can act as a seal (Watts,
1987; Knipe, 1992).
(1) Juxtaposition, in which reservoir sands are
juxtaposed against a low-permeability unit (e.g.,
shale) with a high entry pressure.
(2) Clay smear (i.e., entrainment of clay or shale)
into the fault plane, thereby giving the fault itself a
high entry pressure.
(3) Cataclasis, which is the crushing of sand
grains to produce a fault gouge of finer grained
material, again giving the fault a high capillary
entry pressure.
(4) Diagenesis, when preferential cementation
along an originally permeable fault plane may partially or completely remove porosity, ultimately creating a hydraulic seal.
Juxtaposition seals can be recognized explicitly
by mapping the juxtaposition of units across the
fault plane. Although juxtaposition against tight
lithologies (such as shales) will give the greatest seal
effect, juxtaposition of two sands with different capillary properties will also give rise to a measurable
pressure difference across the fault. This difference
is not due to any fault-zone material; nevertheless, it
may correspond to up to 15 m difference in oil column height between the two sands (Berg, 1975).
To identify or predict sealing by clay smear, cataclasis, or diagenesis requires an ability to relate these
mechanisms to measurable properties of the subsurface (such as lithology and fault displacement). A
further desirable feature of any predictive method
897
898
Figure 1Field example of clay smears separating sandstones from Frechen lignite mines, Germany (Modified
from Weber et al., 1978). Note tapering of clay (black)
away from the source bed and the compound nature of the
clay smear in the fault zone. (No scale on original figure.)
Yielding et al.
899
The exponents m and n can be regarded as additional variables whose values may be justified by
experimental or observational studies. When m =
n, the result is dimensionless and is therefore independent of the units of measurement. As n increases above 1 (as in the CSP equation), thicker source
beds contribute more to the calculation than do
thin beds; i.e., they are weighted by a greater proportion than their increase in thickness.
Lindsay et al. (1993) proposed a shale smear factor to constrain the likelihood of shale smear continuity. Based on their observations of abrasion
smears in a lithified sequence, they define the shale
smear factor (SSF) as (see Figure 2c)
SSF =
Fault throw
Shale layer thickness
(3)
900
Figure 2Smear factor algorithms for estimating likelihood of clay smear on a fault plane. (a) Clay smear potential
(CSP) (Bouvier et al., 1989; Fulljames et al., 1996) given by the square of source-bed thickness divided by smear distance; (b) generalized smear factor, given by source-bed thickness divided by smear distance, with variable exponents;
(c) shale smear factor (SSF) (Lindsay et al., 1993) given by fault throw divided by source-bed thickness. Methods (a)
and (b) model the distance-tapering of shear-type smears, whereas method (c) models the form of abrasion smears.
Yielding et al.
901
) 100%
(4)
The shale thicknesses are measured in a window with a height equal to the throw; therefore,
this window represents the column of rock that
has slid past this point on the fault. The definition
can be extended for cases where the stratigraphic
breakdown is by reservoir zone rather than by
individual beds. In these cases, the net contribution of fine-grained material from each reservoir
zone can be related to the clay content and thickness of the zone. The corresponding equation is
(see Figure 3b)
SGR =
[(
) (
)] 100%
Figure 3Gouge ratio algorithms for estimating likelihood of clay entrainment in the fault gouge zone. The
gouge ratio reflects the proportion of the sealing lithology in the rock interval that has slipped past a given
point on the fault. (a) Calculation for explicit shale/clay
beds in an otherwise shale-free sequence; z is the
thickness of each shale bed. (b) Calculation for a
sequence of reservoir zones; z is the thickness of each
reservoir zone and Vcl is the clay volume fraction in
the zone.
(5)
902
Yielding et al.
903
904
Figure 5Strike projections of fault K in the Nun River field, Nigeria, constructed from figures in Bouvier et al. (1989).
(a) Distribution of footwall lithologies (black = shale, white = sandstone) and proven hydrocarbons in the footwall
sands (green = oil, red = gas). (b) Juxtaposition diagram showing footwall shales in black [as in (a)] and hanging-wall
shales in gray. White areas are sand/sand overlaps. Blue arrows indicate examples of sand/sand contacts where hydrocarbons are trapped at the fault [compare with arrowed locations in (a)]. (c) Fault throw (vertical offset) in meters.
Yielding et al.
905
Figure 5Continued.
906
Figure 6Smear factor attributes calculated for fault K, Nun River field. Attributes are displayed on a strike projection, as in Figure 5. Footwall and hanging-wall shales are shown in black. (a) Clay smear potential (CSP) as defined
by Fulljames et al. (1996); i.e., square of shale bed thickness divided by smear distance. (b) Linear smear factor; i.e.,
shale bed thickness divided by smear distance. Note that CSP gives a greater weighting to thicker shales than does
the linear smear factor (compare fringes around uppermost three shales, labeled c4c, c5c, d1).
Yielding et al.
907
Figure 7Comparison of (a) shale smear factor (SSF) of Lindsay et al. (1993) with (b) shale gouge ratio (SGR) for
fault K, Nun River field. For cases with noncompound smears, these two attributes are expected to have a reciprocal
relationship, and the color bars are set up to reflect this.
A number of wells drilled in different fault compartments allow a calibration of the clay smear calculations for several faults. Because of the heterogeneous nature of the Brent Group reservoir, it was
divided into eight zones (corresponding to the
zones in the reservoir model) and SGR was used
(equation 4). An example of one analyzed fault is
given in Figure 10. This figure shows two strike
projections, the first displaying the juxtaposition of
the Brent reservoir zones, and the second showing
calculated SGR. Note that there is considerable
overlap of the gross reservoir interval (Brent
908
Yielding et al.
909
Figure 9Plots for fault K, Nun River field, illustrating the relationship between across-fault pressure difference and
(a) clay smear potential (CSP), (b) shale gouge ratio (SGR), and (c) shale smear factor (SSF). Each point corresponds to
one grid node on the fault surface (attributes gridded at 20 10 m). Nodes where gas is trapped against the fault (red
in Figure 5a) are shown as crosses, those where oil is trapped against the fault are shown as filled circles. The lines
labeled seal capacity indicate the maximum pressure difference sustainable by a particular value of the attribute.
Figure 10Strike projections of a fault in the Oseberg Syd area (North Sea), from Fristad et al., 1996. (a) The juxtaposition pattern for the Brent reservoir
zones. Upthrown (footwall) zones are shown by colored fill, whereas the downthrown (hanging-wall) zones are shown as labeled black outlines. The reservoir comprises the section between top Brent and base Brent, together with the overlying Heather Sandstone (uppermost orange layer). Hydrocarbon contacts in the footwall are shown in blue dashed lines, and those in the hanging wall as black dashed lines. (b) The pattern of shale gouge ratio (SGR) calculated on the area of reservoir/reservoir overlap. The key area of low gouge ratio occurs near the crest of the overlap zone, indicated by the yellow area at UTM
6693000-6693500; here, SGR drops to about 18%, but the pressure difference across the fault is almost 8 bars (116 psi).
Figure 10Continued.
912
Figure 11Pore-pressure profile across the fault illustrated in Figure 10. The two sides of the fault share a
common aquifer gradient (bottom right). In the hydrocarbon-bearing zone there is a progressive upward
increase in pressure difference across the fault. GOC =
gas-oil contact; OWC = oil-water contact.
Yielding et al.
913
Figure 13Fault-seal calibration for North Sea data set 2. (a) Schematic cross section across the juxtaposed Brent
Group fault blocks. GOC-1 and OWC-1 are the gas-oil and oil-water contacts, respectively, for the hanging wall
(dashed lines); GOC-2 and OWC-2 are the gas-oil and oil-water contacts, respectively, for the footwall (solid lines).
(b) Pore-pressure profiles for the two sides of the fault (dashed and solid lines, respectively). Aquifer and gas pressure gradients are equal on both sides of the fault, but there is seal between the oil legs, which have different thicknesses. (c) Crossplot of shale gouge ratio (SGR) against pressure difference for parts of the fault that separate the
higher pressured oil leg from the other side of the fault.
Calibration Summary
Figures 914 illustrate fault-seal capacity in terms
of one or more of the attributes defined in the previous section. With each of the attributes, there is a
trend or threshold that would allow likely seal
capacity (static pressure difference) to be predicted
from the value of the attribute. In general, similar
914
Yielding et al.
Figure 15Plot of capillary entry and breakthrough pressures for fault gouge samples, using data from Gibson (in
press). Gibsons effective pore-throat radius has been
converted to capillary pressure using Pc = 2 cos /R,
taking = 40 mN/m (Firoozabadi & Ramey, 1988) and cos
= 1. Cataclastic deformation bands consist of fractured
detrital grains, solution deformation bands have welded grain contacts implying pressure solution, and complex deformation bands are cataclastic bands overprinted by pressure solution. Note that the increasing percent
of phyllosilicates corresponds to the increasing minimum capillary pressure.
and typically involves grain breakage and comminution. This results in a significantly reduced grain size
in the fault zone and increased grain packing. The
pore-throat radii are consequently constricted, and
the fault rock can support a pressure difference
because of the increased capillary entry pressure.
Knipe (1992) reviewed microstructural studies of
fault-zone rocks and noted that cataclastic fault
gouge may have pore-throat radii less than 0.001
mm, capable of supporting an oil column height of
up to 300 m. Observations of permeability across
cataclastic slip bands (Antonellini and Aydin, 1994;
Heath et al., 1994) showed that cataclasis may
reduce the permeability by up to four orders of magnitude in high-porosity sandstones, even with only a
few centimeters of displacement. Mandl et al. (1977)
noted in an experimental study that gouge formation
was restricted to initial displacement and did not
increase thereafter. Crawford (1995) also showed
experimentally that initial displacements produce
drastic reductions in permeability.
There is some difficulty relating the microstructural observations to field-scale fault geometries. A
number of studies (e.g., Robertson, 1983; Scholz,
1987; Evans, 1990; Knott, 1994) indicated that
thickness of the zone of deformation around a fault
915
correlates approximately linearly with displacement, such that the thickness is 0.010.1 of the displacement. If permeability continued to decrease
as the fault zone became thicker, measured fault
displacement could be used as a guide to the permeability reduction. However, this seems unlikely,
because the thicker fault zone is composed of anastomosing and en echelon strands that each have a
lower displacement than the total zone. Reservoir
modeling studies of this geometry (Omre et al.,
1994) indicated that the transmissibility of the zone
may be preserved by virtue of flow around individual fault strands; therefore, cataclasis in a faulted
sand probably has a significant effect on seal capacity at low displacements and then only increases
slowly as displacement continues to increase.
Fulljames et al. (1996) suggested that the continuity of the fault zone tends to increase with increasing displacement, and therefore there may be a critical displacement at which the transmissibility
drops to zero (i.e., the fault zone becomes a seal).
This threshold is likely to be dependent on detailed
host-rock lithology (e.g., porosity and subsequent
diagenesis).
A major control on the development of a cataclastic gouge is the magnitude of the stress on the
fault plane during movement (Watts, 1987). Thus,
generation of cataclastic gouge should be favored
by greater depth and reverse and strike-slip faulting
rather than extensional faulting. However, experimental studies (Mandl et al., 1977) showed that
some grain breakage can occur at low overburden
pressures corresponding to only 100 m of burial.
There is a common observation of cataclastic web
structure in DSDP/ODP cores from active margins.
This structure occurs even in poorly consolidated
sands with maximum burial depths of a few hundred meters (e.g., Lucas and Moore, 1986).
Cataclastic deformation bands commonly have
associated with them features of clay smear or diagenesis. Fractures in clay-bearing sandstones show
increased packing and a strong alignment of clay
particles, enhancing their seal potential (e.g., Kent
et al., 1995; Fristad et al., 1996). Zones of cataclasis
are commonly affected by cementation, and quartz
overgrowths develop on the fractured grains,
reducing the pore-throat sizes (or, in extreme cases,
blocking the pore throats), again leading to
enhanced seal potential.
CONCLUSIONS
Observations of clay/shale smears indicate that
their geometries are dependent on the thickness of
the source bed and on the fault throw or smear distance. These relationships have been used to define
a number of fault-surface attributes that describe
916
the likelihood of smears being developed at a particular point on the fault surface. These attributes
are as follows.
(1) Clay smear potential (CSP) (Bouvier et al.,
1989), which is the sum of (thickness2/distance)
for shale beds.
(2) Generalized smear factor, which is the sum of
(thicknessn/distancem) for shale beds, with optional
exponents for both thickness and distance.
(3) Shale smear factor (SSF) (Lindsay et al.,
1993), equal to throw/thickness.
(4) Shale gouge ratio (SGR), which is net
shale/clay percentage in the slipped interval.
Applying these algorithms to fault surfaces in a
variety of data sets shows that they can be related
to the pore-pressure difference supported locally
by the fault; i.e., to the seal capacity. Threshold values of the attributes that differentiate between sealing and nonsealing parts of the fault can be identified. For example, the threshold value for SGR is
about 1520% (cf. 47 for SSF). A progressive
increase in seal capacity is observed with increasing gouge ratio or CSP (or decreasing SSF). The
choice of which particular algorithm to use for
fault-seal prediction in a particular data set should
be based on the available data. Smear factor methods require defining individual shale beds, whereas
the gouge ratio method can also be applied to
zoned reservoir sequences.
Ideally, any study that uses such attributes in
fault-seal prediction needs to be locally calibrated
by appropriate pressure data, as described for the
examples in this paper. However, even without
such calibration in a particular study, useful results
can still be obtained. For example, in an exploration context it may be possible to rank a series of
fault-bound prospects on the basis of their relative
values of a fault-seal attribute, although the local
relationship between attribute and seal capacity
may be unknown.
There is a matter-of-fact tendency among geologists to refer to faults either as sealing or nonsealing. This is an oversimplification, because the combination of lithologic and throw variation ensures
that the fault rock will be heterogeneous. The
methods described in this paper predict that a single fault may seal over some regions of its surface
and leak over others.
REFERENCES CITED
Antonellini, M., and A. Aydin, 1994, Effect of faulting on fluid flow
on porous sandstones: petrophysical properties: AAPG
Bulletin, v. 78, p. 355377.
Arch, J., and A. J. Maltman, 1990, Anisotropic permeability and tortuosity in deformed wet sediments: Journal of Geophysical
Research, v. 95, p. 90359045.
Bentley, M. R., and J. J. Barry, 1991, Representation of fault sealing
in a reservoir simulation: Cormorant Block IV, UK North Sea:
Yielding et al.
917
Tim Needham
Tim Needham received his B.Sc.
degree in geology from Imperial
College, London (1981) and a
Ph.D. from the University of Leeds
(1984). After postdoctoral research
at Durham University and a lectureship in structural geology at
Goldsmiths College, London, he
joined Robertson Research International in 1988. He joined Badley
Earth Sciences as a structural geologist in 1992. His recent interests have centered on fault
seal analysis, fault populations, and analysis of natural
fracture systems.