Uncertainties associated with fault sealing

analysis
1

Jonny Hesthammer1 and Haakon Fossen2

Statoil, N-5020 Bergen, Norway (e-mail: jonhe@statoil.com)

Department of Geology, University of Bergen, Allegt. 41, N-5007 Bergen,
Norway (e-mail: haakon.fossen@geol.uib.no)
ABSTRACT: Recent advances in understanding how faults

restrict fluid flow in sandstone reservoirs have led to improved

models for reservoir simulation. Thevertheless, there are still
many uncertainty factors that can render even the most
detailed simulation model useless. On a detailed scale, these
uncertainties include variations in lateral continuity of faults,
properties and thickness of fault zones, and the influence of
deformation bands within and outside damage zones.
Subseismic features such as small-scale relay zones, drag
features and frequency and distribution of small faults around
the fault zone further decrease the confidence level of
simulation modelling results. Detailed analyses of seismic and
well data from the Gullfaks Field, Thorthern Thorth Sea, have
helped
understand
the
detailed
structural
reservoir
characteristics. The results from these analyses can, in many
cases, be used as input to further enhance models for reservoir
simulation in order to increase the validation of the models.
Furthermore, the studies carried out on the Gullfaks Field
demonstrate that a sound approach to knowledge management
for increased oil recovery based on fault seal analysis requires
sharing of gained knowledge from many oil and gas fields
rather than monopolizing information that cannot be fully
utilized by studies from a single field.
KThYWORDS: Gullfaks Field, fault plane, uncertainty, sealing

characteristic, reservoir communication

INTRODUCTIO
N
The last decade has seen rapid growth in our
understanding of how faults aect fluid flow in
oil and gas reservoirs (Allan 1989; Bouvier et al.
1989; Bentley & Barry 1991; Antonellini & Aydin
1994; Gibson 1994, 1998; Knott et al. 1996;
Lopez & Smith
1996; Childs et al. 1997; Fristad et al. 1997;
Fulljames et al. 1997;
Knipe 1997; Lia et al. 1997; Yielding et al. 1997,
1999; Crawford 1998; Foxford et al. 1998; Knai
& Knipe 1998; Manzocchi et al. 1998, 1999;
Ottesen Ellevset et al. 1998; Walsh et al. 1998a,
b; Fossen & Hesthammer 1998b; Hesthammer
1999b). These advances have helped to establish
sound methods for calculat- ing fault sealing
potential as a function of rock properties and
fault characteristics (Knipe 1997; Yielding et al.
1997; Manzocchi et al. 1999).
These methods may, however, fail if the
reservoir character- istics are not fully
understood. For example, known fault rock
properties (Yielding et al. 1997; Manzocchi et al.
1999) can be combined with fault juxtaposition

diagrams (Allan 1989; Knipe 1997) to evaluate

the sealing capacity of faults. The input is
generally based on displacement estimates from
seismic interpretation, known rock properties and
the general relation- ship between fault zone
thickness and displacement. The eect of
subseismic faults, local drag and width of damage
zone (not fault zone; see Fig. 1) is not considered.
These factors can drastically change the eects of
fluid flow across the barriers
and must be
considered in a thorough
investigation.
This paper focuses on how the already
established methods for calculating fault sealing
potential and modelling fluid flow can benefit
from newly acquired knowledge based on studies
of dipmeter data, core data and seismic data
from the Gullfaks

Field. This includes information on fault

frequency and characteristics of drag zones and
damage zones. This integrated approach should
lead to a fuller and better understanding of
how and to what extent fluid flow is restricted
by faults.

ThSTABLISHThD
MThTHOD
AND
UNCThRTAINTIThS
Several works have discussed how fault seal
potential can be calculated when the lithologic
reservoir properties and fault properties are
known (Yielding et al. 1997; Knipe 1997). As a
next logical step in understanding fluid flow,
Manzocchi et al. (1999) introduced a method for
using fault transmissibility multipliers for flow
simulation, thereby allowing detailed investigations into the complex topic of reservoir
simulation. The procedure is based on the
Petroleum Geoscience, Vol. 6 2000, pp. 3745
Society, London

recognition that a fault displaces reservoir units

of various quality. In addition, the fault zone
itself restricts fluid flow.
The displacement of flow units can be
evaluated by creating fault juxtaposition
diagrams along the fault (Bentley & Barry 1991;
Knipe 1997). As the dierent stratigraphic units
are associated with a variety of permeability,
porosity and shale contents, the flow across the
fault will vary. As such, permeability profiles can
be constructed along the
fault.
The permeability of the fault zone in sandstone
reservoirs is mainly dependent on the shale
content and fault zone thick- ness. Manzocchi et
al. (1999) use the Shale Gouge Ratio (SGR)
method established by Yielding et al. (1997) to
determine the shale content of the faulted
sequence. This method calculates
the
proportion of phyllosilicate that is displaced past
a particu- lar point on a fault. Since the lowest
SGR value will have the
1354-0793/00/$15.00 g2000 EAGE/Geological

3
8

J. Hesthammer and
H. Fossen

Fig. 1. (a) Well log correlation

diagram showing the location
of a fault associated with 6 m
of missing section in well
34/10-A-15. (b) Fracture
frequency diagram from well
34/10-A-15. A damage zone
with abundant deformation
bands is associated with the 6
m fault identified from well
log correlation data. (c) Core
photograph from the interval
around the 6 m large fault.
The thin fault zone is clearly
identified, as are abundant
deformation bands near the
fault zone.

lowest capillary entry pressure, the method can

be used to identify which points are likely to leak
fluids and which points will represent barriers to
fluid flow. Generally, a SGR greater than 1520%
results in a membrane seal (Watts 1987). The
fault zone thickness is mainly dependent upon
the displacement along the fault (Robertson
1983; Hull 1988; Childs et al. 1997; Walsh et al.
1998a; Manzocchi et al. 1999), although some
studies also indicate a relationship between fault
zone thickness and lithology (Childs et al. 1997;
Knott et al. 1996). As such,
a simple
relationship between fault zone thickness and
displacement may be used as input in
modelling studies.
Manzocchi et al. (1999) aimed to predict fault
zone properties by the use of a simple algorithm.
They recognized that the method depends on
several assumptions and approximations and
clearly state that the model needs calibration
against
dynamic reservoir data, although
considerable
uncertainty
will
always
be
associated with the fault transmissibility simply
due to the natural unpredictability of fault zone
structure and shale content. For instance,
Foxford et al. (1998) concluded from studies of
the Moab Fault in Utah that the fault structure
cannot be predicted over distances greater than
ten metres. Lia
et al. (1997) considered the
fault transmissibility to be the largest factor of
uncertainty related to reserves estimated for the
Veslefrikk Field (located along the eastern flank
of the Viking Graben). Based on the uncertainties
related to fault zone properties and thickness,
Manzocchi et al. (1999) expect at least two
orders
of
magnitude
variation
in
fault
permeability at any particular SGR.

The uncertainty factors related to estimates of

restrictions on fluid flow across faults can be
summarized as follows.
(1) The rapid spatial changes in continuity and
geometry
along faults are dicult or
impossible to predict and will greatly aect
the predictions of fluid flow in a faulted
reservoir.
(2) Faulted rock can have many dierent
characteristics, such as fault gouge, clay
smear,
breccia,
cement,
lenses
of
undeformed rock and deformation bands.
The dierent types of fault rock determine to
a large extent how eectively the fault will
restrict fluid flow. Diagenetic processes (for
instance related to burial depth) may greatly
aect the fault rock properties.

Fault sealing

(3) Although a simple general relationship

may
analysis
be established for fault zone thickness
estimates, huge
variations
will
exist
depending on deformation mechanism and
rheology. For instance, the dip of the fault
surface, shale content of reservoir rocks
and the degree of consolidation at the time
of deformation can greatly aect the
resulting fault zone thickness.
(4) Shale content within a fault zone is a
function of the rheology (ductility) of the
shale which is dependent on water content
and porosity (a function of overburden) at
the time of deformation.
(5) Deformation around faults in sandstones is
not only related to the fault zone itself.
Abundant deformation bands exist in a
damage zone outside the fault zone.
Deformation bands can drastically restrict
fluid flow and must be considered in
reservoir simulation and estimates of fluid
flow restriction across faults.
(6) Deformation
bands
exist
sporadically
outside damage zones and will aect fluid
flow patterns depending on their geometry
and how much they reduce permeability.

They should, therefore, be incorporated 9

into
simulation models.
(7) Relay structures on a subseismic scale can
cause pathways for fluid flow across a fault
that would otherwise be considered
sealing.
(8) Seismic resolution is generally too poor to
define whether
a seismically interpreted
fault is a single structure or consists of
numerous smaller faults. This uncertainty
may enhance or reduce estimates of fluid
flow across fault barriers and can cause
large errors in the prediction of fault
sealing capacity.
(9) Many faults are associated with drag on a
subseismic scale. The eect is somewhat
similar to having several small faults rather
than one large structure in that fault
juxtaposition diagrams based on seismic
interpretation
will be wrong. However,
whereas many small faults will represent
several barriers, a drag zone where the
deformation
mechanism
is
by
reorganization of individual grains (common
in loosely consolidated sandstones) will
only cause a minor decrease in porosity and
permeability.

In general, lateral discontinuities (1) will always

represent a serious uncertainty for simulation of
fluid flow in reservoirs.
The uncertainties
related to fault rock properties (2), average fault
zone thickness (3), shale rheology (4) and
damage zones
(5) can be reduced by examining any available
core data (see below). The geometry of isolated
deformation bands outside damage zones (6) can
only be partially extracted from core data. The
lateral geometries (interaction with other
deformation
bands and lengthdisplacement
relationships) can only be resolved by studying
outcrop
data
and
will
represent
large
uncertainties depending on how closely the
outcrop data match those within the reservoir. A
similar uncertainty is related to relay structures
(7) as these cannot, in general, be studied using
data collected from within the reservoir or from
seismic data (although pressure and well history
data combined with attribute maps may, in
some cases, be helpful; see fig. 5b in
Hesthammer & Fossen 1997a). Whether a major
displacement structure consists of one or more
faults (8) can to some extent be resolved by
detailed analyses of well log correlation data.
Finally, dipmeter data and core data can help
understand the abundance and characteristics of
drag zones (9).
In the following section, we will discuss how
the un- certainties outlined above (with the
exception of lateral discon- tinuities) can be
reduced by detailed analyses of outcrop and
subsurface data. This includes field analogue
data, seismic data and abundant well data (well
log correlaton data, dipmeter data and core
data). The information has been collected from
field studies in Utah and investigations of
available data from the Gullfaks Field, Thorthern
Thorth Sea, and presented in more detail
elsewhere (Fossen & Hesthammer 1997, 1998a,
b, in press; Hesthammer & Fossen 1997a, b,
1998; Hesthammer 1999b).
RThDUCING THTh UNCThRTAINTY
Fault zone properties, fault zone thickness
and shale content
When studying the fault zone characteristics
from areas other than the reservoir to be
modelled, it is crucial that the deformation
mechanisms are comparable. A common mistake
is to assume that outcrop data can be directly
compared to subsurface oil and gas reservoirs.
This is only true if the lithology, deformation
mechanism and degree of consolidation are
comparable. Thot uncommonly, deformation in
outcrop analogues occurred after the rocks were
consolidated. In addition, the rocks have later
been uplifted and subjected to erosion (causing
fracturing). In contrast, many of the sub- surface
reservoirs (e.g. Thorth Sea fields) were deformed
during or immediately after deposition when the
sediments were only loosely consolidated.
Consequently, the fault zones are less aected by
cataclasis, the shales were more ductile and the
deformation was not necessarily restricted to
narrow zones. These dierences will drastically
aect the fault zone properties (whether a fault
zone is brecciated, cemented, anastomosing

etc.), fault zone thickness and shale content (more

ductile shale leads to more continuous shale
smears).
If good estimates of uncertainties related to
fault zone properties are to be obtained, the
geoscientist needs to incor- porate all available
core data (and FMS/FMI data if available).
However, this is not an easy task since it is very
hard to obtain good core data through a fault
zone. On the Gullfaks Field, 18 faults have been
cored within a total cored interval of 6 km. All of
these are identified by several metres of missing
section (as identified from well log correlation)
and the presence of a damage zone with abundant
deformation bands (Fig. 1). How- ever, only for 3
of the 18 faults has the fault zone itself been

Fig. 2. Principal sketch of fault structure,

consisting of a central
fault zone of intense
deformation, and an enveloping zone of
microfaults or deformation bands. Deformation
bands may also occur outside the damage zone as
single or aggregate structures.

preserved during the core operations. One of

these faults is a cemented breccia. As such, it is
not possible from these data alone to obtain
reliable estimates of fault zone thickness as a
function of displacement. Furthermore, this
scarceness of data does not allow for a
comparison with onshore field data to evaluate
if they are comparable. The Gullfaks Field is
one of the most heavily faulted reservoirs in the
Thorth Sea and clearly demonstrates the
diculty
in
understanding
uncertainties
related to fault zone properties. In order to
resolve this problem, it is necessary to
compile
available
data
from many more
oshore fields that have similar lithologies and
deformation history.

Damage zone and the importance of

deformation bands
Studies of faults in sandstones in Utah and the
Gullfaks Field demonstrate that abundant
deformation bands exist in a narrow zone around
the fault (Figs 1 and 2). This zone is generally
less than a few tens of metres wide and contains
several tens to a
few hundred deformation
bands (Hesthammer 1999b). The deformation
bands are dierent from other faults in that they
lack a discrete slip surface (Fossen &
Hesthammer
1998b).
Instead,
they
are
associated with a strain hardening process in
which the individual grains eventually crack and
become more densely packed, thereby reducing
the permeability by up to four orders of
magnitude. The exact amount of permeability
reduc- tion depends on the amount of cataclasis
and phyllosilicate content (Antonellini & Aydin
1994; Statoil 1997). In Utah, the deformation
bands are associated with much cataclasis that
significantly reduces permeability (typically by
three orders of magnitude).
On the Gullfaks Field, only minor cataclasis is
present (Hesthammer 1999b) and the reduction
in permeability is controlled by the phyllosilicate
content (Statoil 1997). When the phyllosilicate
content
increases
past
1820%,
or
if
phyllosilicate layers exceed 2040% per metre,
the deformation structures on the Gullfaks Field
are dominated by a phyllosilicate framework
which reduces the permeability by 2 or 3 orders
of magnitude (Statoil 1997). When the
phyllosilicate content exceeds 40%, phyllosilicate
smear is the dominant process, and the permeability is reduced to less than 1 D.
Deformation
bands
in
clean
sandstones
(containing less than a few per cent phyllosilicates) on the Gullfaks Field are characterized
by a denser packing of sand grains and have only
very limited eect on fluid flow. In comparison,
recent studies of deformation bands from

Gullfaks Sr demonstrate that these bands are

associated with abundant quartz dissolution
which seriously lowers the porosity and
permeability within the bands and thus restricts
flow across them. The reason for the dierences
observed between the Gullfaks Field and
Gullfaks Sr is related to the dierent depths at
which the reservoir rocks are located. Whereas
the Gullfaks Field reservoir rocks are generally
located at 18002500 m depth, reservoir rocks
on Gullfaks Sr are commonly situated below
3000 m depth. At such depths, the temperature
exceeds 120C, thus allowing for accelerated
quartz dissolution. Since deformation bands tend
to favour localized fluid flow along the bands,
such zones will experience more dissolution of
quartz grains than the surrounding rocks.
Obviously, the failure to incorporate deformation
bands and their physical properties into fault
sealing analysis studies would not reveal any
dier- ences between fluid flow across faults on
the Gullfaks Field and Gullfaks Sr, an error that
has already resulted in serious complications
associated with production from the Statfjord
Formation on Gullfaks Sr.
The width of the deformation bands on the
Gullfaks Field is a function of deformation band
type, grain size and mineralogy (Fig. 3). Thus by
knowing the phyllosilicate and quartz content of
the reservoir rocks (obtained from well logs) as
well as the grain size (obtained from core data),
the amount of permeability reduction and
width of deformation bands can be found. This
can be combined with information on the number
of deformation bands associated with a fault to
provide a sound statistical input for reservoir
simulation. Deformation bands
tend to form
prior to the development of a discrete slip
surface (Antonellini & Aydin 1994; Fossen &
Hesthammer 1997, 1998b). As such, there is no
clear relationship between the width of the
damage zone and the displacement across the
fault zone as observed on the Gullfaks Field (Fig.
4) (there may be some indications that faults
with missing sections larger than
15 m are
associated with a wider damage zone; possibly
related to the more complex geometries of larger
faults). Studies from the Gullfaks Field show that
the average width of a single deformation band
is 1.4 mm (Hesthammer 1999b). Assuming that
all faults identified by well log correlation
represent a single slip surface, the average width
of a damage zone is somewhat more than 24 m
and the average number of deformation bands
within a damage zone is slightly higher than 179
(Table 1). Although this type of statistic can be
used for general simulaton purposes on the
Gullfaks Field (the numbers may be fieldspecific), they should not be applied to single
cases as the individual variations are large (Table
1).
Core studies from the Gullfaks Field
demonstrate that approximately 70% of the
deformation bands occur within damage zones
where the bands are interconnected in complex
networking zones (Hesthammer 1999b). The
remaining 30%, located outside damage zones,
commonly occur as single deformation bands or
linked by soft-link or hard-link structures to only
a few other bands (Fossen & Hesthammer

1998b). Since a single deformation band can be

associated with a permeability reduction of as
much as four orders of magnitude and thereby
significantly reduce fluid flow, the geometry of
these bands is important to understand.
Furthermore, studies from Utah (Fossen &
Hesthammer 1997, 1998b) show that deformation
bands in consolidated sandstones have a
displacementlength relationship that is dierent
from faults with discrete slip surfaces (Fig. 5). The
deformation bands are much longer than faults
with discrete slip surfaces for the same
displacement. Although this relationship is clear
from studies in Utah, it is unclear whether the
same relationship applies to deformation bands on
the Gullfaks Field (which developed in poorly
consolidated sandstones). In fact, recent studies
from Morocco

Fig. 3. (a) Width of deformation bands versus type.

Deformation bands associated with cataclasis are
wider than those that are only aected by
disaggregation structures. (b) A plot of grain size
versus width of deformation bands show that the
width increases with increasing grain size. (c) Plot
of the width of deformation bands with respect to
the main constituent. Quartz-rich clean sandstones
are generally associated with wider deformation
bands than sandstones rich in phyllosilicates.

(Wibberley et al. 1999) may indicate that

deformation bands in loose sand behave like
ordinary faults (Fig. 5). Until more data are
available, it is not possible to use this
information to accurately simulate the eect a
single deformation band has on fluid flow. Some
general statistics can be implemented by
considering the percentage of deformation bands
located out- side damage zones. In addition,
studies of fracture frequency diagrams from the
Gullfaks Field show that deformation bands

Fig. 4. Plot of missing section

(as identified from detailed
well log correlation) associated
with faults versus the width of
the damage zone. There is no
clear relationship to suggest
that the width of damage zones
increases systematically with
increasing oset.
However, there are large
uncertainties related to the
larger-scale faults, and it is
possible that faults with more
than approximately 15 m of
missing section have wider
damage zones than smaller
faults. This may be due to the
more complex geometry of
larger-scale faults.

located outside damage zones occur as 15

bands within a narrow zone of less than 0.5 m
width.
Subseismic structures
Studies of outcrop data demonstrate that faults
commonly exhibit relay structures (Cartwright et
al. 1995) rather than single

continuous surfaces. The relay structure may be

soft-linked, in which the faults aect each other
without physically touching, or hard-linked, in
which the faults are connected across the relay
zone. Both types of structure will typically
improve rather than restrict communication. A
soft-linked
relay
structure
provides
a
communication path for fluids across an
otherwise

Table 1. (a) General statistics for deformation bands. See main text for discussion. (b) Information
related to damage zones associated with larger-scale faults that have developed discrete slip
surfaces
(a)

Formation
All
Heather
Tarbert
Thess
Etive
Rannoch
Drake
Cook
Amundsen
Statfjord
Lunde

Deformati
on bands
total
(number)

Width
(avera
ge)
(mm)

Displacem
ent
(average)
(mm)

4824
151
669
970
201
1624
46
299
54
679
131

1.42
ThA
1.65
1.02
2.21
1.05
2.51
1.02
ThA
1.69
2.13

8.61
ThA
11.18
9.24
5.00
5.43
ThA
14.11
ThA
8.35
13.50

Damage
zone
From (mMD)

Damage
zone
To
(mMD)

DZ
width
(m)

1912
1848
1885
2121
2281
2577
2425
2223
2883
2083
2363

>1933
1853
1892
2127
2286
2581
2457
>2285
2913
2086
2442

>21
5
7
6
5
4
32
>62
30
3
79

(b)
Well

34/10-5
34/10-A5H
34/10-A5H
34/10-A8
34/10-A14
34/10-A15
34/10-A16
34/10-B1
34/10-B12
34/10-C1
34/10-C3

Deformation
band
in sDZ
>63
4
106
238
23
109
236
>789
210
84
188

Missing
section
(m
)
15
6.5
?
5
10
11
20
45
69
14
18

34/10-C5
34/10-C14
34/10-C14
Sum
DZ, damage
zone.

<3115
3650
<3543

3159
3673
3563

>44
23
>20
>341

>272
25
>155
>250
2

?
8
?
221.5

the Gullfaks Field demonstrates that 119 (42%) of

the faults occur within intervals less than 100 m
wide (along the wellbore path). These faults are
located too close to each other to be identified
from seismic interpretation. As many as 194
(69%) of the faults are located less than 300 m
apart (along the wellbore path). Assuming a
lateral resolution of less than 300 m, it can be
inferred from this study that more than 69% of

Fig. 5. Displacementlength diagram for faults and

deformation
bands.
The
plot
shows
that
deformation bands in consolidated sandstones from
Utah (Fossen & Hesthammer 1997) are longer than
faults with discrete slip surfaces, whereas
deformation bands
in unconsolidated sandstones
from Morocco (Wibberley et al. 1999) display the
same slope as faults with discrete slip surfaces.
The sources for other previously published fault
data are cited in Schlische et al. (1996).

sealing structure, whereas a hard-linked

structure increases the likelihood of preserving a
sand-to-sand contact across the fault (due to a
smaller oset associated with each fault). A
typical problem with seismic interpretation of oil
and gas fields is that the relay structures are
below seismic resolution and therefore remain
undetected by the seismic interpreter. On the
Gullfaks Field, limits in seismic resolution do not
in general allow definition of faults located less
than a few hundred metres apart (Hesthammer
& Henden in press). Thus, unless the fault
strands within a relay structure are placed
suciently far apart, the faults will likely be
interpreted as a single, continuous
surface.
However, the presence of kinks and sudden
changes
in fault strike may often indicate the
location of a relay structure and should be
considered when studying fault seal potential for
a specific area. For full field reservoir simulation,
the uncertainty related to the presence of relay
structures will always exist and will drastically
reduce the reliability of the analyses.
A related problem is the ability to distinguish
whether a fault zone consists of more than one
fault surface. Commonly on the Gullfaks Field,
faults interpreted as a single structure from
seismic data are shown to consist of several
smaller
faults
based on detailed well log
correlation. Although this will always remain
an uncertainty in areas not penetrated by wells,
statistical analyses can be used for general
simulation purposes.
A study of 280 faults
identified from detailed well log correlation on

the faults identified from seismic interpretation

on the Gullfaks Field will consist of more than a
single fault surface (the true number is likely to
be somewhat higher since not all faults in a
fault zone are likely to be penetrated by a single
well). Unless this is accounted for, fault seal
analyses based on seismic interpretation alone
will be erroneous in most
cases.
Figure 6a shows that faults on the Gullfaks
Field are more abundant at shallower reservoir
levels (Tarbert, Thess and Rannoch formations).
In addition, Fig. 6b shows that the displacement
increases with depth. There is no clear relationship between numbers of faults and lithology
(Drake,
Amundsen and parts of Thess
formations consist mainly of shale whereas the
other formations contain mainly sandstone).
This important information can be used to
model the eect of fault sealing capacity with
respect to formation and depth. However,
drilling of more than 180 wells on the Gullfaks
Field clearly demonstrates that the local
variations are great. Whereas a main fault can
be proved to consist of a single fault surface by
the drilling of one well, drilling of another well
through the same fault a few hundred metres
away may reveal that the fault consists of 35
smaller faults.
The eect of drag of bedding towards a fault
will also drastically influence fluid flow paths
across a fault structure (Fig. 7). A study of 23
km of dipmeter data from the Gullfaks Field
demonstrates that as much as 60% of all faults
on the
field are associated with drag of
bedding on a scale that is below seismic
resolution (Hesthammer & Fossen, 1998).
Further- more, the dierence in missing section
as identified from well log correlation and the
total oset outside the zone aected by drag
can be as much as one order of magnitude (Fig.
8). Obviously, this important information must
be considered when modelling fluid flow in a
reservoir. The individual variations are too large
to allow a systematic use of these data
(at

least based on present analyses). The sensitivity

to the uncertainties can be estimated by the use
of Allan diagrams (Allan 1989; Knipe 1997).
Further important information from analysis of
dipmeter data is that drag is only associated
with northsouth trending faults on the Gullfaks
Field (Fig. 9a). There is no clear relationship
between displacement and the width of the drag
zone (Fig. 9b). This may be explained by drag
being developed prior to the establishment of a
distinct fault slip surface (Hesthammer & Fossen
1998). Once a distinct slip surface has
developed, further deformation will occur along
this weak zone, and the drag zone becomes
fossilized. The average width of the detected
drag zones is 60 m in the hanging wall and 29 m
in the footwall (measured along the wellbore
path). There is no clear relationship between the
width of the drag zone and lithology or depth,
probably because the sandstones were only
loosely consolidated when deformation occurred
in the late Jurassic (Hesthammer & Fossen
1998). Faults at deeper stratigraphic levels are
less aected by drag than faults at shallower
reservoir levels (Fig. 10). This is probably
because deeper strati- graphic levels were
somewhat more consolidated at the time of
deformation. Although the results from analyses
of dipmeter data are not always conclusive, it
should be possible to enhance existing models
for fluid flow simulations by incorporating the
general information obtained from the many
detailed analyses carried out on the Gullfaks
Field.
Several studies have focused on the abundance
of faults in sandstone reservoirs close to or
below the limits of seismic resolution (Jones &
Knipe 1996; Hesthammer & Fossen 1997a, b;
Hesthammer 1998, 1999a). In general, the fault
population in the Gullfaks Field follows a powerlaw distribu- tion down to approximately 510 m
displacement
(Fossen
&
Rrnes
1996).
Furthermore, there appears to be a gap in
fault

Fig. 6. (a) Faults at shallower

reservoir levels (Brent Group)
are more abundant than at
deeper stratigraphic levels.
(b) The oset associated with
faults increases with depth. See
main text for discussion.

Fig. 7.
Due to the eect of local drag on a
subseismic scale, the real communication path may
be very dierent to that estimated from the use of
fault juxtaposition diagrams. This must be
considered when evaluating fault seal potential.

frequency between 10 cm and 510 m (Fossen &

Hesthammer in press). Based on seismic data
alone, it is easy to overinterpret the presence
and locations of small-scale faults (Hesthammer
1999a). Thus, extreme caution is advised if
seismic interpret- ation alone is to provide the
input to models of reservoir performance. Failing
to do so can cause serious misinterpret- ations
which may, in the worst case, lead to an
erroneous plan for field development.

DISCUSSION AND CONCLUSIONS

Several advances in understanding fluid flow
across faults have led to large improvements
in models for reservoir simulations. There are,
however, many important factors governing
fault sealing potential
that
are
not
incorporated into the existing models. The
above analyses demonstrate how the existing
models for estimating fault seal potential and
understanding fluid flow in sandstone reservoirs
can benefit from detailed analyses of seismic
and well data from oil and
gas fields.
The use of core data can help reduce
uncertainties related to estimation of fault zone
thickness and
properties.
Similarly,
core
analysis helps understand the eect of
deformation bands on reservoir simulation,
although the eect of lateral variations cannot
be revealed by analyses of core data alone.

Dipmeter
data help identify the presence and
characteristics of drag zones, whereas well log
correlation can reveal if the fault structure
consists of one large- or several smaller-scale
faults.
By combining seismic data, well log
correlation data, dipmeter data and core data,
detailed information can be obtained about
structural
reservoir
characteristics.
This
information can then be used to enhance the
models for simulating fluid flow in reservoirs.
Although many of the results from the Gullfaks
Field can be used for general evaluation of
reservoir behaviour, extreme caution must be
used when applying the data to individual case
studies (such as evaluation of fault seal potential
across a single fault for input to decisions on
whether to drill a well or not). In order to
evaluate the generality of the results, it is
necessary to quality control the established
models against dynamic data

Fig. 8. Studies of dipmeter

data from the Gullfaks Field
show that the dierence
between total oset (i.e. the
oset outside the zone aected
by drag) and missing section
can be drastically dierent, as
shown for a fault identified in
well 34/10-C-3. See main text
for detailed discussion.

obtained from wells. On the Gullfaks Field,

where data from more than 115 km of drilled
reservoir from 180 wells exist, such quality
control is possible, but the large amount of data
and high structural complexity require much
future work to fully utilize all the available
information.
Many of the analyses carried out on the
Gullfaks Field are
still associated with large
uncertainties. The results from the analyses may
not be applicable to other fields. In order to
further enhance our understanding of reservoir
properties, it is necessary to compile data from
other fields in a manner similar to that carried
out on the Gullfaks Field. This demands an

Fig. 9. (a) Orientation analysis shows that mainly

northsouth trending faults on the Gullfaks Field are
associated with drag. (b) There is no clear
relationship between missing section associated with
a fault and the interval aected by drag. This is
probably because drag mainly developed prior to the
development of a distinct slip surface (see main text).
However, the drag zone is generally wider in the
hanging wall than in the footwall.

approach to knowledge management where

oil companies are willing to share information
in order to optimize oil recovery. Competition,
as such, should not be associated with
Fig. 10. A plot of formation versus number of
faults associated with drag per kilometre shows

that faults at shallower stratigraphic levels (Brent

Group) are more often associated with drag than
faults at deeper stratigraphic levels. This is
probably because the deeper strata was more
consolidated at the time of deformation.

monopolization of routine data (seismic, core and

standard well log data) but rather how that
shared information is utilized.
The authors would like to thank Statoil, Thorsk
Hydro and Saga Petroleum for permission to
publish the article. The manuscript has benefitted
from comments by Graham Yielding and an
anonymous reviewer.

RThFThRThNCT
hS
ALLATh, U. S. 1989. Model for hydrocarbon migration and
entrapment.
American Association of Petroleum Geologists Bulletin,
73, 803811.
AThTOThELLIThI, A. & AYDITh, A. 1994. Eect of faulting
on fluid flow in porous sandstone: petrophysical
properties.
American
Association
of
Petroleum
Geologists Bulletin, 78, 181201.
BEThTLEY, M. R. & BARRY, J. J. 1991. Representation of
fault sealing in a reservoir simulation: Cormorant block
IV UK Thorth Sea. In: Proceedings of the 66th Annual
Technical Conference & Exhibition of the Society of
Petroleum Engineers, Dallas Texas, 119126.
BOUVIER, J. D., KAARS-SIJPESTEIJTh, C. H.,
KLUESThER, D. F.,
OThYEJEKWE, C. C. & VATh DER PAL, R. C. 1989.
Three-dimensional seismic interpretation and fault
sealing investigations, Thun River Field, Thigeria.
American Association of Petroleum Geologists Bulletin,
73, 13971414.
CARTWRIGHT, J. A., TRUDGILL, B. D. & MAThSFIELD, C.
S. 1995.
Fault growth by segment linkage: an explanation for
scatter in maximum displacement and trace length data
from the Canyonlands Grabens of SE Utah. Journal of
Structural Geology, 17, 13191326.
CHILDS, C., WALSH, J. J. & WATTERSOTh, J. 1997.
Complexity in fault zones and implications for fault seal
prediction. In: Mller-Pedersen, P. & Koestler, A. G. (eds)
Hydrocarbon seals: Importance for Exploration and
Production. Special Publication of the Thorwegian
Petroleum Society, 7, 6172.
CRAWFORD, B. R. 1998. Experimental fault sealing: shear
band per- meability dependency on cataclastic fault
gouge characteristics. In: Coward,
M. P., Johnson, H. & Daltaban, T. S. (eds) Structural
Geology in Reservoir Characterization. Geological
Society, London, Special Publications, 127, 2744.
FOSSETh, H. & HESTHAMMER, J. 1997. Deformation
bands and their significance in porous sandstone
reservoirs. First Break, 16, 2125.
& 1998a. Structural geology of the Gullfaks
Field. In: Coward,
M. P., Johnson, H. & Daltaban, T. S. (eds) Structural
Geology in Reservoir Characterization. Geological
Society, London, Special Publications, 127, 231261.
& 1998b. Geometric analysis and scaling
relations of deformation bands in porous sandstone.
Journal of Structural Geology, 19, 14791493.
& in press. Possible absence of small faults in the
Gullfaks Field, northern Thorth Sea: implications for
downscaling of faults in some porous sandstones.
Journal of Structural Geology, 21.
& RRThES, A. 1996. Properties of fault populations
in the Gullfaks Field, northern Thorth Sea. Journal of
Structural Geology, 18, 179190.
FOXFORD, K. A., WALSH, J. J., WATTERSOTh, J.,
GARDETh, I. R.,
GUSCOTT, S. C. & BURLEY, S. D. 1998. Structure and
content of the Moan Fault Zone, Utah, USA and its
implications for fault seal predictions. In: Jones, G.,
Fisher, Q. J. & Knipe, R. J. (eds) Faulting, Fault Sealing
and Fluid Flow in Hydrocarbon Reservoirs. Geological
Society, London, Special Publications, 147, 87103.
FRISTAD, T., GROTH, A., YIELDIThG, G. &
FREEMATh, B. 1997.
Quantitative fault seal prediction A case study from
the Oseberg Syd. In: Mller-Pedersen, P. & Koestler, A.
G. (eds) Hydrocarbon seals: Importance for Exploration

and Production. Special Publication of the Thorwegian

Petroleum Society, 7, 107124.
FULLJAMES, J. R., ZIJERVELD, J. J. & FRAThSSETh, C. M.
W. 1997. Fault
seal processes: systematic analysis of fault seals over
geological and production time scales. In: MllerPedersen, P. & Koestler, A. G. (eds) Hydrocarbon seals:
Importance for Exploration and Production. Special
Publication of the Thorwegian Petroleum Society, 7, 51
59.
GIBSOTh, R. G. 1994. Fault-zone seals in siliclastic strata of
the Columbus Basin, oshore Trinidad. American
Association of Petroleum Geologists Bulletin, 78, 1372
1385.
1998. Physical character and fluid-flow properties of
sandstone derived fault gouge. In: Coward, M. P.,
Johnson, H. & Daltaban, T. S. (eds) Structural Geology in
Reservoir Characterization. Geological Society, London,
Special Publications, 127, 8397.
HESTHAMMER, J. 1998. Evaluation of the timedip,
correlation
and
coherence
maps
for
structural
interpretation of seismic data. First Break, 16, 151167.

 1999a. Improving seismic data for detailed

structural interpretation. The Leading Edge, 18, 226
247.
1999b. Fault geometry and internal fault block
deformation in the Gullfaks region, northern North
Sea. PhD thesis, University of Bergen, Thorway.
& FOSSETh, H. 1997a. Seismic attribute analysis in
structural inter- pretation of the Gullfaks Field,
northern Thorth Sea. Petroleum Geoscience, 3, 1326.
& 1997b. The influence of seismic noise in
structural inter- pretation of seismic attribute maps.
First Break, 15, 209219.
& 1998. The use of dipmeter data to constrain
the structural geology of the Gullfaks Field, northern
Thorth Sea. Marine and Petroleum Geology, 15, 549
573.
& HEThDETh, J. In press. Closing the gap between
theory and practice in seismic interpretation of smallscale faults. Petroleum Geoscience.
HULL, J. 1988. Thickness-displacement relationships for
deformation zones.
Journal of Structural Geology, 10, 431435.
JOThES, G. & KThIPE, R. J. 1996. Seismic attribute maps;
application to structural interpretations and fault seal
analysis in the Thorth Sea Basin. First Break, 14, 449
461.
KThAI, T. A. & KThIPE, R. J. 1998. The impacts of faults
on fluid flow in the Heidrun Field. In: Jones, G., Fisher,
Q. J. & Knipe, R. J. (eds) Faulting, Fault Sealing and
Fluid Flow in Hydrocarbon Reservoirs. Geological
Society, London, Special Publications, 147, 269282.
KThIPE, R. J. 1997. Juxtaposition and seal diagrams to
help analyze fault seals in hydrocarbon reservoirs.
American Association of Petroleum Geologists Bulletin,
81, 187195.
KThOTT, S. D., BEACH, A., BROCKBAThD, P. J.,
BROWTh, J. L.,
MCCALLUM, J. E. & WELDOTh, A. I. 1996. Spatial and
mechanical controls on normal fault populations.
Journal of Structural Geology, 18, 359372.
LIA, O., OMRE, H., TJELMELAThD, H., HOLDETh, L. &
EGELAThD, T.
1997. Uncertainties in reservoir production forecasts.
American Association of Petroleum Geologists Bulletin,
81, 775801.
LOPEZ, D. L. & SMITH, L. 1996. Fluid flow in fault
zones:
influence
of
hydraulic
anisotropy
and
heterogeneity on the fluid flow and heat transfer
regime. Water Resources Research, 32, 32273235.
MAThZOCCHI, T., RIThGROSE, P. S. & UThDERHILL, J. R.
1998. Flow
through fault systems in high porosity sandstones. In:
Coward, M. P., Johnson, H. & Daltaban, T. S. (eds)

Structural Geology in Reservoir Characteri- zation.

Geological Society, London, Special Publications, 127,
6582.
, WALSH, J. J., ThELL, P. & YIELDIThG, G. 1999. Fault
transmissibility multipliers for flow simulation models.
Petroleum Geoscience, 5, 5363.
OTTESETh ELLEVSET, S., KThIPE, R. J., OLSETh, T. S.,
FISHER, Q. T. &
JOThES, G. 1998. Fault controlled communication in the
Sleipner Vest Field, Thorwegian Continental Shelf:
detailed, quantitative input for reser- voir simulation
and well planning. In: Jones, G., Fisher, Q. J. & Knipe,
R. J. (eds) Faulting, Fault Sealing and Fluid Flow in
Hydrocarbon Reservoirs. Geological Society, London,
Special Publications, 147, 283297.
ROBERTSOTh, E. C. 1983. Relationships of fault
displacement to gouge and breccia thickness. Institute
of Mining Engineers, 35, 14261432.
SCHLISCHE, R. W., YOUThG, S. S., ACKERMAThTh, R. V. &
GUPTA, A.
1996. Geometry and scaling relations of a population of
very small rift-related normal faults. Geology, 24, 683
686.
STATOIL. 1997. Faulting and fault sealing in the Gullfaks
Field. Fisher,
Q. J., Knipe, R. J., Condclie, D. & Jones, R. M.
unpublished report.
WALSH, J. J., WATTERSOTh, J., HEATH, A. E. & CHILDS, C.
1998a.
Representation and scaling of faults in fluid flow models.
Petroleum Geoscience, 4, 241251.
, , , GILLESPIE, P. A. & CHILDS, C. 1998b.
Assessment on
the eects of sub-seismic faults on bulk permeabilities of
reservoir sequences. In: Coward, M. P., Johnson, H. &
Daltaban, T. S. (eds) Structural Geology in Reservoir
Characterization. Geological Society, London, Special
Publications, 127, 99114.
WATTS, Th. L. 1987. Theoretical aspects of cap-rock and
fault seals for single- and two-phase hydrocarbon
columns. Marine and Petroleum Geology, 4, 274307.
WIBBERLEY, C. A. J., PETIT, J.-P. & RIVES, T. 1999.
Mechanics of high
displacement gradient faulting prior to lithification.
Journal of Structural Geology, 21, 251257.
YIELDIThG, G., FREEMATh, B. & ThEEDHAM, D. T. 1997.
Quantitative
fault seal prediction. American Association of Geologists
Bulletin, 81, 897917.
, VERLAThD, J. A. & BYBERG, G. 1999.
Characterization of fault zones for reservoir modeling:
An example from the Gullfaks Field, northern Thorth
Sea. American Association of Geologists Bulletin, 83,
925951.

Received 30 April 1999; revised typescript accepted 27 September 1999.