Downloaded from http://pgc.lyellcollection.

org/ by guest on September 14, 2015

Characterization of fault zones in the Gullfaks Field for reservoir modelling

G. Y I E L D I N G , 1 J. A. 0 V E R L A N D 2 and G. B Y B E R G 2 3

1
Badley Earth Sciences Ltd, North Beck Lane, Hundleby, Spilsby, Lines PE23 5NB
(e-mail: graham@badleys.co.uk)
2
Norwegian Petroleum Directorate, Prof. Olav Hanssensvei 10, N-4001 Stavanger, Norway
3
Present address: Statoil, 4035 Stavanger, Norway

Abstract: A fault-seal study has been performed for faults around the A platform of the Gullfaks Field,
Northern North Sea. The operator's depth grids and isochore maps were used to build a FAPS database.
Together with petrophysical logs, these data were used to compute the distribution of shale gouge ratio
(SGR: fault-zone % shale) over each fault surface.
Pressure data from exploration and production wells have been projected onto the modelled fault
surfaces. Where well data permit, across-fault pressure drops at particular time instants during production
are displayed. These dynamic pressure drops provide a guide to the permeability of the fault zones once
flow has started.
The calculated fault parameters (throw and SGR) can be converted to other parameters such as fault-
zone thickness and permeability. These can be used to derive fault transmissibility modifiers for reservoir
simulations. The calculated properties are compatible with the recorded pressure history and tracer move-
ment between wells.

The objective of this study was to investigate the sealing
potential of the internal faults in the Gullfaks Field (Norwegian
Block 34/10), both with regard to a static model (differences in
original fluid contacts) and with regard to a dynamic model
(production history). The linkage between fault properties and
differences in fluid contacts has been clearly demonstrated in
several studies in recent years (e.g. Fristad et al. 1996; Yielding
et al. 1997). It has been more difficult to find a quantitative
expression for fault behaviour when exposed to the sudden
changes in pressure caused by production (e.g. Jev et al. 1993).
In this study we analyse the areas of probable leakage in the
faults by comparing measured pressure differences over faults
with the sealing properties (expressed as shale content in the
fault zone) of those faults. An ultimate aim for the study is to be
able to produce a quantitative expression, in grid format, of
the fault-zone permeabilities. Based on the permeability values,
fault transmissibility modifiers for simulation models can be
calculated.
Reservoir properties
Erichsen et al. (1987) and Petterson et al. (1990) describe the
reservoir properties on the Gullfaks Field. (The operator's
petrophysical interpretation has been used throughout this
study.) The present study focuses on the Brent Group,
although the Cook and Statfjord Formations are also
important reservoir units.
The Brent Group reservoir consists of the sandstone units
within the Tarbert, Ness, Etive and Rannoch formations (the
Broom Formation is thin and considered non-reservoir).
Tarbert consists of massive, homogeneous and highly perme-
able (3-10 D) sandstone units with a few shale, coal and car-
bonate layers. Ness/lower Tarbert consists of thin sandstone
units interbedded with shale and coal layers, which act as
vertical flow and pressure barriers. The massive sandstone of
the Etive Formation has excellent reservoir properties (2-7 D).
Reservoir properties degrade from top to bottom of the
Rannoch Formation (2-0.050 D).

Need for better understanding and characterization of

Gullfaks Field
Structural setting
Internally the Gullfaks fault block is highly faulted (Fig. 1).
Fossen & Hesthammer (1997) divide the Gullfaks fault block
into two structurally distinct subareas: a major domino system,
and an eastern horst complex, separated by an accommoda-
tion zone. The two subareas show significant differences with
respect to fault geometry, rotation and internal block defor-
mation. The major internal faults on the Gullfaks block are
E-dipping with strike approximately N-S. Kinematic studies
show that these faults are dip-slip extensional structures
(Fossen & Hesthammer 1997). E-W-striking minor faults,
both N- and S-dipping, divide the major domino blocks
into smaller fault compartments. The operator's names for
the internal fault compartments (El, E2 etc.) are used in
this study.
fault properties
Production from the Gullfaks Field commenced in December
1986. In the planning and early production phase, it was real-
ized that the extent of faulting on the field would have a major
impact on the strategy for reservoir management (Petterson
et al. 1990). Pressure measurements, both with repeat forma-
tion tester (RFT) in new wells and with permanent downhole
gauges, and the use of non-radioactive tracers, were considered
important to measure the effects of the production and the
injection across the faults. From these data it was realized
already in 1988 that some degree of pressure communication
existed between the separate fault compartments.
On the Gullfaks Field surplus gas, which has not been sold
because of restrictions in transportation capacity, has been
reinjected for storage. Attempts have been made to use some
of the surplus gas for WAG (water alternating gas) injection to

YIELDING, G., 0VERLAND, J. A. & BYBERG, G. 1999. Characterization of fault zones in the Gullfaks Field for reservoir modelling. In: FLEET, A. J. &
BOLDY, S. A. R. (eds) Petroleum Geology of Northwest Europe: Proceedings of the 5th Conference, 1177-1185. ©Petroleum Geology '86 Ltd.
Published by the Geological Society, London.
 Downloaded from http://pgc.lyellcollection.org/ by guest on September 14, 2015
1178 G. YIELDING ET AL.

(a)

(b)
A G2/G3 Fault
B
1.6
G3 \ G2 H1

Fig. 1. (a) Top Brent structure map of the study area around the A-platform, showing fault compartments and well locations, redrawn from a
map provided by Statoil (field operator). The fault highlighted in red is the example fault discussed in detail ('G2/G3 fault'). The dashed line labelled
A-B shows the line of the section shown in (b). (b) Cross-section along the line marked A-B in (a). E3, F4, G3, G2 and HI refer to the fault
compartments identified in the map view. The G2/G3 fault is highlighted in red.

improve the displacement of attic oil. In both of these cases,
the injected gas has been produced back from unexpected
well locations, often within a relatively short time. This move-
ment of injected gas has been documented by the application of
tracers (Kleven et al. 1995). The Norwegian Petroleum Directo-
rate was interested in evaluating the possibility of applying gas
injection on a larger scale as an integrated part of the reservoir
management on the Gullfaks Field. One obvious requirement
to be able to achieve this goal is a better understanding of the
flow and pressure patterns in the reservoir. During this study
special attention was given to explaining the observed move-
ments of gas.
Focus of study
The study was focused on the part of the Brent reservoir that is
drilled and produced from the A-platform (Gullfaks A-area)
(Fig. 1). The reason for this is three-fold:
(1) The Brent Group reservoir contains the major part of the
oil resources on the Gullfaks Field. This has resulted in a
large number of production wells penetrating this reser-
voir, giving a dense well database.
(2) Production on Gullfaks started in the A-area, so this area
would contain the earliest, most undisturbed, and there-
fore probably most reliable, pressure measurements.
(3) The quality of the acquired 3D seismic data deteriorates
over the eastern part of the field, because of gas leakage.
Over the western part of the field, where the presence of
Upper Jurassic shales has prevented gas leakage, clearer
seismic data allow credible fault identification. The 3D-
seismic line spacing is 12.5x 12.5m, and the effective
resolution for fault offsets is believed to be about 20-30 m
(Hesthammer & Fossen 1997). The detailed structural data
collected in the production wells are incorporated into the
operator's seismic interpretation, resulting in detailed and
accurate fault mapping.
 Downloaded from http://pgc.lyellcollection.org/ by guest on September 14, 2015
FAULT-ZONE CHARACTERIZATION, GULLFAKS
Down thrown*
horizon
Fig. 2. Sequence of diagrams showing (a) modelled fault grid with seismically-mapped horizons; (b) interpolation of detailed reservoir zonation
between the mapped horizons; and (c) reservoir-reservoir overlap area.
Although the study was focused on the Brent reservoir,
parameters for the overlying Upper Jurassic Shale (Viking
Group) and the underlying Dunlin Group were added to be
able to calculate fault properties where these units juxtapose
the Brent reservoir.
Most of the Brent Group reservoir in the Gullfaks Field has
a common oil-water contact of 1947 m below msl. The excep-
tions are the westernmost fault blocks, E2 and E3 (Fig. 1),
which have a shallower oil-water contact. In addition, a small
gas cap is observed in fault block Gl (Fig. 1). For estimating
the fault seal potential during production, high-quality pres-
sure measurements (RFT) taken in open boreholes are used.
Pressure measurements taken early in the production history,
before the start of water injection, are expected to show a more
unique relation to the fault properties, than the complex
pressure situation experienced after years of production and
injection. Particularly suitable calibration points are pressure
measurements in 'undrained' fault blocks that showed pressure
drops caused by the production in adjacent blocks.
Methodology
A more complete description of the general methodology used
by FAPS software for fault analysis and fault-seal analysis is
given by Needham et al. (1996), Freeman et al. (1997) and
Yielding et al. (1997).
The principal steps by which the fault-seal analyses were
constructed from the map and well database were as follows:
(i) Use the fault traces at all mapped horizons to define the
positions of the fault surfaces in depth. Seven horizons
had been mapped by the operator (Base Cretaceous, Top
Brent, Top Ness, Top Rannoch, Base Brent, Top Cook,
Top Statfjord).
(ii) Model each fault surface as a three-dimensional grid, with
principal axes along-strike and down-dip, and grid-cell
size 25 m x 25 m (see schematic illustration in Fig. 2).
(iii) Define the detailed reservoir zonation (from well data), by
interpolating zone boundaries between the mapped hori-
zons at the fault surface (also illustrated in Fig. 2). A total
of 23 reservoir zones were used within the Brent Group.
(iv) Input relevant RFT (pressure) measurements to each
reservoir, to display the pore-pressure field in the fault-
block. Where pressure data have been collected on both
sides of a fault, the pore-pressure field can be constructed
for both fault-blocks (i.e. both walls of the fault). The
difference between these distributions is the pressure differ-
ence across the fault zone (assuming that the pressure field
within each compartment is relatively uniform, i.e. all
pressure changes occur across the faults).
1179
Fault plane
Viewing direction
(v) Assign shale-volume (' F shale ') data to the reservoir zones,
by deriving zonal averages from the Fshaie logs. Shale
content is a critical parameter in the calculation of fault
seal capacity (Yielding et al. 1997; Gibson 1998). We use
an attribute called the shale gouge ratio (SGR) which is
simply the percentage of shale or clay in the rock interval
that has slipped past any point on the fault (see Fig. 3).
Core and outcrop data suggest that this parameter is a
good guide to the proportion of phyllosilicates in the fault
gouge (Gibson 1997; Foxford et al. 1998), which in turn
is a major influence on the capillary and permeability
properties of the fault zone. In the fault analyses, SGR is
calculated over the fault surface on a refined grid of
5 m x 5 m (i.e. five times the resolution of the basic fault
grid described earlier).
Permeability measurements on fault gouge samples (Gibson
1997; Ottesen Ellevset et al. 1998) permit a first-order cor-
relation between phyllosilicate content (~SGR) and fault-zone
permeability. For example, a phyllosilicate content of < 15% in
the fault zone is characteristic of cataclasites with permeability
c. 0.5 mD, whereas archetypal clay smears have phyllosilicates
>40% and permeability <0.003mD. Fault-related diagenesis/
cementation is not significant in the Gullfaks reservoir, and
hence fault-gouge phyllosilicate content is expected to be
the main control on across-fault permeability. The SGR
distribution on the modelled fault surface can therefore be
used as a prediction of fault-zone permeability.
In turn, the fault-zone permeability can be used to derive a
transmissibility modifier for each grid node on the fault grid.
The transmissibility modifier is simply the ratio between the
transmissibilities across the fault with and without the effect
of the fault-zone material. To calculate the cell-to-cell trans-
missibility in the absence of fault-zone material requires the
permeabilities of the juxtaposed reservoir units, and the
dimensions of the grid cells in the reservoir model. Calculat-
ing the transmissibility with fault-zone material between the
cells additionally requires the fault-zone permeability and
its thickness. Here the fault-zone permeability is derived
from the SGR (as described earlier). The fault-zone thickness
can be input as a constant (e.g. based on fault penetrations
in well cores) or can be considered as a simple function of
the local fault displacement (typically one hundredth, e.g.
Evans 1990).
In this paper only the mapped faults in the field are
described. However, the same methodology can in principle be
extended to synthetic faults of small displacement, to investi-
gate the possible effect of sub-seismic faulting (see Yielding
et al. in press).
 Downloaded from http://pgc.lyellcollection.org/ by guest on September 14, 2015

1180 G. YIELDING ET AL.

coded areas are those parts of the fault planes where there is
Brent-Brent overlap (Fig. 2c). A wide overlap area on the map
indicates a relatively small fault displacement, whereas a
narrow (or zero) overlap area corresponds to larger displace-
ment. The analysis of each fault uses zone thickness and KShaie
data from the wells adjacent to that fault. The SGR scale can
\ throw = t be considered as representing static seal capacity, or being
inversely related to fault-zone permeability. It can be seen
1
from Fig. 5 that faults with small displacements (wide areas

-•! throw window, t

on map) have a heterogeneous SGR distribution. Faults with
'Shale Gc )uge Ratio' \ V.,,Az
cl5' s
5 large displacements (narrow areas) have a more uniform (and
at point (3n fault is: \ | moderately high) SGR distribution.
V AZ
Qr"iP — V A / A -,\ / t- „ 4 0 0 0 / A cl4' 4 The next section focuses on one of the analysed fault
OUK — L{ V . .L\£-) 11 A 1 \J\J /O \ V
c,3'AZ3 segments, to show in detail how the fault surfaces can be
(i.e. % clay in slipped interval) VAZ2 characterized geometrically and in terms of their fluid flow
V^ V
cn-AzT
behaviour.

Fig. 3. Definition of shale gouge ratio (SGR). For a sequence of

reservoir zones with specified Fshaie (or Vc\ay) and thickness, SGR
equals the net shale/clay content in the rock interval that has slipped
past any point on the fault. SGR is considered to be a good estimate
of the likely phyllosilicate content of the fault gouge on that part
of the fault.
Results of fault analyses
A total of 23 fault segments were analysed by the methods
outlined in the previous section. The analysed faults form an
interconnected network that defines a number of separated
fault compartments (Fig. 4). Fault throw in the Brent Group is
typically in the range 0-200 m, which is less than the total
Brent Group thickness. Only one of the faults analysed (in the
southwest of the area) has sufficient displacement to offset
the Brent Group completely (Fig. 4). Fault-zone thickness for
these faults will typically be about one-hundredth of the local
displacement (e.g. Evans 1990).
Figure 5 shows, in map view, the distribution of SGR on all
23 analysed fault segments in the Gullfaks A area. The colour-
G2/G3 fault
This example fault lies between fault blocks G2 and G3, which
originally had the same oil-water contact. Therefore, on a
geological timescale it probably did not constitute a 'sealing'
fault. However, the two fault-blocks underwent differential
depletion once production had started, implying that the fault
was a barrier to flow on the production timescale.
Figure 6 is a perspective view of the fault surface, colour-
coded by the fault throw. The throw reaches about 50 m on the
northern part of the fault but is much less in the south (c. 10-
20 m). At both ends the fault links onto other faults (Fig. 1),
i.e. there are no lateral tips. The fault was penetrated near its
southern end by well A-19 (unfortunately the fault plane was
not cored).
Figure 7 shows the calculated SGR in the reservoir-reservoir
overlap areas using the lowest F sha i e values from adjacent wells.
Table 1 summarizes FShaie values from the wells, and the min.
and max. values used in the calculations. A prominent feature
of the plot is the area of lower SGR values in the Lower Brent,

Fig. 4. Map view of the analysed faults showing the throw (vertical offset) distribution at the level of Top Ness. The large fault in the southwest has
throws reaching 350 m; on all other faults the Top Ness throw is <200m and does not completely offset the Brent Group. Fault-zone thicknesses
are likely to be about 1/100 of the fault displacement (Evans 1990).
 Downloaded from http://pgc.lyellcollection.org/ by guest on September 14, 2015
FAULT-ZONE CHARACTERIZATION, GULLFAKS 1181

Gouge
ratio(%)
0
10
20

0.5 km
30
40
50 •
Fig. 5. Map view of the analysed faults showing SGR in the Brent-Brent overlap zones. The calculation for each fault uses the iow-case' Fshaie
values from adjacent wells. The mapped width of each overlap zone depends upon the fault dip and amount of offset; for the largest fault
in the southwest there is total Brent offset and therefore no overlap zone (cf. Fig. 4).

especially on the southern half of the fault. The small offset of
the clean Etive Formation is the main cause of this.
In Fig. 8, the SGR distribution has been used to calculate
a transmissibility modifier at all reservoir-reservoir overlaps.
A transformation was used from gouge ratio to fault-zone
permeability, and another transformation from fault throw to
fault-zone thickness. In combination with the permeability
values in the reservoir zones, these transformations allow a
comparison of the transmissibility from footwall to hanging
wall, with and without the fault-zone, at all points on the fault
surface. Expressed as a ratio, this comparison is by definition
the transmissibility modifier.
Because a number of variables affect the transmissibility
modifier, it does not necessarily have a simple relationship with
SGR. In general, for a given SGR, juxtapositions of highly
permeable reservoirs have a lower transmissibility modifier,
because the fault zone has a more significant effect on the
transmissibility. Conversely, when poorly permeable reservoirs
are juxtaposed the fault zone has relatively less effect and the
transmissibility modifier is higher (i.e. nearer to 1).
The first well in block G3 was Well A-8 in mid-1987; RFT
data showed only very minor depletion from earlier produc-
tion to the northeast. A-8 production began in June 1987. Well
A-10 was drilled into block G2 in July 1987 and found 7.5 bar

" WelhMsP
^YrmVWX)^ ArYV\VvA\

^ V \ \--"\ Vf\ y ^ VvS^niVt-H

\\ y fe^N
—
i \ \ V\~v3r\ 3^WV \ \ \\\vvSCS
Throw (m)

§§|f| u

20

40;

\ \_-v-\ \ \ W-'\ \ T 60

Fig. 6. G2/G3 fault, between fault-blocks G2 and G3 (see Figs 1, 4 & 5 for location). The fault surface is colour-coded by throw (vertical
displacement). The total length of the fault is about 1.2 km and each grid cell is 25 m x 25 m.
 Downloaded from http://pgc.lyellcollection.org/ by guest on September 14, 2015

1182 G. YIELDING ET AL.

Fig. 7. G2/G3 fault, showing low-case distribution of SGR (or fault-zone %-shale) in reservoir overlap zones (using 'Min' values from Table 1).
Note the area of low SGR on the deeper part of the Brent-Brent overlap (adjacent to the Etive Fm). The fault grid is shown in light grey. The thin
black lines show reservoir zone boundaries in the upthrown (G3) side of the fault. The thin coloured (orange/yellow) lines show reservoir
zones on the down thrown side (G2).

depletion in the Rannoch/Etive. This must have been caused
either by A-8 production in G 3 , or by ongoing production to
the east in block H I , or both. Well A-11 ( R F T Sept. 1987) in
G 3 found that d r a w d o w n by A-8 production had now reached
24 bar.
In late 1988, R F T measurements were taken in rapid suc-
cession in Well A-18 (block G3) a n d Well A-19 (G2/G3). Well
A-19 penetrated the G 2 / G 3 fault, passing from Etive in the
down thrown block G 2 to R a n n o c h in u p t h r o w n block G 3 .
Between these units, Well A-19 found a 6 bar pressure differ-
ence. In every other well, the Rannoch/Etive forms a continu-
ous pressure c o m p a r t m e n t because of the lack of shaly breaks.
Well A-19 is therefore directly sampling the pressure d r o p at
one point on the fault surface.
Wells A-18 a n d A-19 together provide pressure profiles for
b o t h sides of the fault for late 1988. In the footwall (G3)
depletion was greatest in the Rannoch/Etive ( > 15 bar) and in
the upper Ness (produced in A-14). In the hanging wall (G2)

Table 1. Vshale data for G2/G3 fault

Mapped Zone Thickness (m) Fshaie values from wells Fshaie values used

A30 A19 A8 A26 Min. Average Max.

Top Brent t3 21 0.193 eroded eroded 0.193 0.19 0.193

t2b2 15 0.168 0.168 0.17 0.168
t2bl 15 0.190 0.327 0.189 0.189 0.24 0.327
t2a 11 0.039 0.185 0.092 0.054 0.039 0.09 0.185
tic 5 0.511 0.319 0.459 0.371 0.319 0.42 0.511
tlb 12 0.275 0.236 0.301 0.313 0.236 0.28 0.313
tla 10 0.360 0.410 0.311 0.328 0.311 0.35 0.410
Top Ness n3d 11 0.298 0.314 0.335 0.301 0.298 0.31 0.335
n3c 8 0.239 0.303 0.287 0.271 0.239 0.28 0.303
n3b 8 0.280 0.309 0.324 0.339 0.280 0.31 0.339
n3a 13 0.220 0.272 0.472 0.282 0.220 0.31 0.472
n2c 14 0.321 0.348 0.195 0.152 0.152 0.25 0.348
n2b2 14 0.428 0.366 0.405 0.336 0.336 0.38 0.428
n2bl 10 0.166 0.235 0.318 0.217 0.166 0.23 0.318
n2a 6 (F) 0.243 0.416 0.292 0.243 0.32 0.416
nlc 2 0.060 0.468 0.171 0.040 0.040 0.18 0.468
nib 9 0.483 0.159 0.529 0.546 0.159 0.43 0.546
nla 2 0.148 0.015 (F) 0.021 0.015 0.06 0.148
etive 34 0.154 0.113 0.144 0.119 0.113 0.13 0.154
Top Rannoch r3 25 0.291 0.254 0.222 0.219 0.219 0.25 0.291
r2 21 0.217 0.231 0.252 0.253 0.217 0.24 0.253
rl 14 0.360 0.319 0.336 0.336 0.319 0.34 0.360
broom 7 0.653 0.585 0.576 0.560 0.560 0.59 0.653
Base Brent drake 0.535 0.523 0.558 0.533 0.523 0.54 0.558

Fshaie data used in the analysis of the G2/G3 fault. Data from 4 wells were used to derive min., max. and average FShaie values for each reservoir zone.
(F) — faulted out.
 Downloaded from http://pgc.lyellcollection.org/ by guest on September 14, 2015
FAULT-ZONE CHARACTERIZATION, GULLFAKS 1183

\r\xfo^\Y\y\AA
Fig. 8. G2/G3 fault, showing transmissibility modifier in overlap zone. Values are mainly <0.1, indicating that the fault will reduce transmissibility
between juxtaposed grid cells in G2 and G3 by a factor of 10 or more.

depletion was more uniform (Rannoch and Ness production
from A-10). The difference between these depletion patterns at
the fault surface is illustrated in Fig. 9. In general block G3 is
more depleted than block G2 (area shown as blue on figure) as
there were more producing wells in G3. However, the central
part of the Brent-Brent overlap zone shows G2 (hanging wall)
depletion locally greater than G3 (red), depending on the
precise juxtaposition of Ness zones. The complexity of this plot
is illustrative of the complex nature of the fault response to
production.
One year after Wells A-18 and A-19 were drilled, another
pair of wells (A-16A, A-26) were drilled in blocks G2 and G3,
allowing a second measurement of pressures in both blocks.
By this time, the Ness and Tarbert pressures in G2 and G3 had
dropped substantially because of production, but pressures in
the Lower Brent had recovered to near-original as the rate of
water injection in several wells exceeded the rate of production.
Figure 10 shows the calculated pressure drop at the fault
surface at this time. Juxtaposition of downthrown depleted
Ness against upthrown recovered Lower Brent gives a very
large pressure drop (>40 bar) in the central part of the overlap
area. This contrasts with the lower part of the overlap area
(Lower Brent against Lower Brent) where the pressure drop is
only about 5 bar from footwall to hanging wall, and also with
the upper part of the fault (downthrown Tarbert) where the
footwall was slightly more depleted.
The pressure observations summarized in Figs 9 and 10
highlight a number of points. Firstly, the changing production/
injection pattern can produce not only changes in the across-
fault pressure drops, but complete reversals in the direction of
the pressure drop. If these pressure drops are associated with
fluid flow, then the direction of fluid flow across the fault will
have reversed between late 1988 and late 1989. Secondly, the
direction of pressure drop (and, therefore, fluid flow) can be in
different directions on different parts of the same fault, at the
same time.

Fig. 9. G2/G3 fault, showing dynamic pressure drop between juxtaposed reservoirs in late 1988, based on RFT data from Wells A-18 and A-19.
Blue areas have a higher pressure in the downthrown side and so fluid movement will be from the downthrown (G2) to upthrown (G3) side.
Well A-19 measured a 6-bar pressure drop across the fault.
 Downloaded from http://pgc.lyellcollection.org/ by guest on September 14, 2015
1184 G. YIELDING ET AL.

Fig. 10. G2/G3 fault, showing dynamic pressure drop between juxtaposed reservoirs in late 1989, based on RFT data from Wells A-16A and A-26.
Note the complete change in across-fault pressure difference relative to that shown in Fig. 9. On the lower part of the fault the driving pressure is now
from the upthrown to the downthrown side (red areas).

In March 1991, non-radioactive tracer was injected into Well
A-11 in block G3 during a phase of gas injection (Kleven et al.
1995). Tracer injection was into the Rannoch (Lower Brent).
There was very rapid detection of tracer across the fault in
block G2 (lower Ness in Well A-19) after only 1-2 months. This
contrasted with detection in other wells in injection block G3,
where the tracer did not appear for more than three years.
These observations imply that the dominant flow direction in
the G3 Lower Brent at Well A-11 was across the fault into
block G2. This flow route can be easily understood in terms
of the juxtapositions and gouge ratio on the southern part of
the fault, together with the prevailing pressure regime. The
juxtaposition geometry is shown simplified in the cross-section
in Fig. 11. Injection of tracer was into the Rannoch of Well
A-11, and tracer would therefore have risen into the overly-
ing Etive since the Rannoch/Etive behaves as a continuous
sand (no shale barriers). The Etive is juxtaposed against lower-
most Ness at the fault, and the SGR plot of Fig. 7 suggests
that SGR is relatively low on this area of the fault (i.e. fault
permeability relatively high). Flow across the fault into the
lower Ness is therefore possible. Pressure measurements in
early 1991 suggest that the pattern seen in Fig. 10 was continu-
ing, i.e. strong pressure drive from footwall Etive into depleted
hanging wall Ness.
Well A-10, the Rannoch producer closest to the A-11
Rannoch injector, did not detect tracer until a year after
injection. Figure 11 shows that the route from A-11 Rannoch
to A-10 Rannoch would be a difficult one, passing down at the
fault offset at a region where the SGR is rising (fault perme-
ability decreasing). Thus, most of the tracer moving into block
G2 would have by-passed the A-10 Rannoch section and is
likely to have been higher in the sequence.
fault surfaces provide a general guide to the permeability
of the fault zones once across-fault flow has started. How-
ever, an overriding control on dynamic pressure drop is
the reservoir management strategy, i.e. the degree to which
pressure drawdown is imposed in particular fault-blocks
in this area of poor aquifer support. Flow direction across
a fault can reverse direction as the production/injection
pattern changes. Moreover, flow can be in opposite direc-
tions on different parts of the same fault at the same time.
(3) The fault-surface modelling also includes predictive
algorithms (SGR) that can be used to derive fault-zone
permeabilities and transmissibility modifiers for reservoir
simulation studies. Areas of higher SGR on the fault sur-
face correspond to lower fault-zone permeability. On the
Gullfaks Field the areas of high and low SGR are
compatible with the recorded pressure history and tracer
movement between wells. The relationship between gouge
ratio and transmissibility modifier is more complex,
A-11
Injection

Conclusions
(1) Fault-surface modelling of the faults in the Gullfaks A
area has provided a framework in which to visualize the
reservoir zone juxtapositions at faults..
(2) Pressure data from appraisal and development wells can
be incorporated into the fault-surface modelling to provide
displays of the across-fault pressure differences before and
during production. The dynamic pressure drops across
Fig. 11. Cross-section of the southern part of the G2/G3 fault between
Wells A-11 and A-10. On the fault trace, red indicates high SGR, green
indicates low SGR (taken from Fig. 7). Tracer injected into the
Rannoch in Well A-11 travelled rapidly to the Lower Ness at
producing Well A-19, across an area of fault plane with low SGR
(relatively high permeability).
 Downloaded from http://pgc.lyellcollection.org/ by guest on September 14, 2015
FAULT-ZONE CHARACTERIZATION, GULLFAKS
depending also on the permeabilities of the reservoir units
separated by the fault, and the fault-zone thickness. The
lowest transmissibility modifiers (i.e. largest degradation in
across-fault transmissibility) occur where a thick fault-zone
of high SGR separates highly permeable reservoirs.
We are grateful to members of the PL050/050B licence group, headed
by Statoil as operator, for permission to publish this study, though the
views expressed here are ours and not necessarily those of Statoil or
the licence partners. We are also grateful to Eva Halland of Norwegian
Petroleum Directorate for the impetus to publish this work. We thank
W. de Boer, T. Reston and A. Hurst for their comments on the first
version of this manuscript.
References
ERICHSEN, T., HELLE, M., HENDEN, J. & ROGNEBAKKE, A. 1987.
Gullfaks. In: SPENCER, A. M. (ed.) Geology of the Norwegian Oil
and Gas Field. Graham & Trotman, London, 273-286.
EVANS, J. P. 1990. Thickness-displacement relationships for fault
zones. Journal of Structural Geology, 12, 1061-1065.
FOSSEN, H. & HESTHAMMER, J. 1997. Structural geology of the
Gullfaks Field, northern North Sea. In: COWARDS, M. P. et al.
(eds) Structural Geology in Reservoir Characterization. Geological
Society, London, 127, 231-261.
FOXFORD, K. A., WALSH, J. J., WATTERSON, J., GARDEN, I. R.,
GUSCOTT, S. C. & BURLEY, S. D. 1998. Structure and content of
the Moab fault zone, Utah, U.S.A. In: JONES, G., FISHER, Q. &
KNIPE, R. (eds) Faulting, Fault Sealing and Fluid Flow in
Hydrocarbon Reservoirs. Geological Society, London, Special
Publications, 147.
FREEMAN, B., YIELDING, G., NEEDHAM, D. T. & BADLEY, M. E. 1998.
Fault seal prediction: the gouge ratio method. In: JOHNSON, H. D.,
& COWARD, M. P. (eds) Structural Geology in Reservoir
Characterization and Field Development. Geological Society,
London, Special Publications, 127, 19-25.
FRISTAD, T., GROTH, A., YIELDING, G. & FREEMAN, B. 1996. Quanti-
tative fault seal prediction - a case study from Oseberg Syd area.
In: Hydrocarbon Seals - Importance for Exploration and Produc-
tion (conference abstracts). Norwegian Petroleum Society, Oslo.
1185
GIBSON, R. G. 1998. Physical character and fluid-flow properties of
sandstone-derived fault gouge. In: JOHNSON, H. D. & COWARD,
M. P. (eds) Structural Geology in Reservoir Characterization and
Field Development. Geological Society, London, Special Publica-
tions, 127, 83-97.
HESTHAMMER, J. & FOSSEN, H. 1997. Seismic attribute analysis in
structural interpretation of the Gullfaks Field, northern North
Sea. Petroleum Geoscience, 3, 13-26.
JEV, B. I., KAARS-SIJPERSTEIJN, C H., PETERS, M. P. A. M., WATTS,
N. L. & WILKIE, J. T. 1993. Akaso Field, Nigeria: use of
integrated 3D seismic, fault-slicing, clay smearing and RFT
pressure data on fault trapping and dynamic leakage. American
Association of Petroleum Geologists Bulletin, 11, 1389-1404.
KLEVEN, R., HOVRING, O., OPDAL, S. T., BJORNSTAD, T., DUGSTAD,
0 . & HUNDERE, I. A. 1995. Non-radioactive tracing of injection
gas in reservoirs. Society of Petroleum Engineers, reprint 35651.
NEEDHAM, D. T., YIELDING, G. & FREEMAN, B. 1996. Analysis of fault
geometry and displacement patterns. In: BUCHANAN, P. G. &
NIEUWLAND, D. A. (eds) Modern Developments in Structural
Interpretation, Validation and Modelling. Geological Society,
London, Special Publications, 99, 189-199.
OTTESEN ELLEVSET, S., KNIPE, R. J., OLSEN, T. S., FISHER, Q. J. &
JONES, G. 1998. Fault controlled communication in the Sleipner
Vest Field, Norwegian Continental Shelf; detailed, quantitative
input for reservoir 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, 283-297.
PETTERSON, O., STORLI, A., LJOSLAND, A. & MASSIE, I. 1990. The
Gullfaks Field: geology and reservoir development. In: BULLER,
A. T., BERG, E., HJELMELAND, O., KLEPPE, J., TORSTETER, O. &
AASEN, J. O. (eds) North Sea Oil and Gas Reservoirs - II. Graham
& Trotman, London, 67-90.
YIELDING, G., FREEMAN, B. & NEEDHAM, D. T. 1997. Quantitative
fault seal prediction. American Association of Petroleum Geolo-
gists Bulletin, 81, 897-917.
, OVERLAND, J. A. & BYBERG, G. in press. Characterisation of
fault zones for reservoir modelling: an example from the Gullfaks
Field, northern North Sea. American Association of Petroleum
Geologists Bulletin.