7 spe-93312-pa-p fractured reservoir characterization using dynamic data in a carbonate field

Fractured Reservoir Characterization
Using Dynamic Data in a Carbonate
Field, Oman
S.I. Ozkaya, Baker Atlas Geoscience, and P.D. Richard, Petroleum Development Oman

Summary development involved waterflooding by means of an inverted
The main objective of this study was to extract fracture data from nine-spot vertical well. Soon after the initiation of the waterflood-
multiple sources and present it in a form suitable for reservoir ing program, early water breakthrough made it clear that the field
simulation in a fractured carbonate field in Oman. Production is by was more faulted and fractured than originally anticipated. The
water injection. A combination of borehole image (BHI) logs and drilling pattern was subsequently changed in 1994–95 to a vertical
openhole logs from horizontal wells revealed that water encroach- line drive oriented parallel to the dominant northwest/southeast-
ment occurs mostly through fracture corridors and appears as sharp trending fault/fracture pattern, following an assessment of 3D seis-
saturation spikes across fracture clusters. Dispersed background mic faults and fractures, fault cutouts, BHI logs, and early pro-
joints have little flow potential because of cementation, lack of duction data (Arnott and Van Wunnik 1996).
connectivity, or small size. Image logs indicate that fracture cor- The fractures and faults of the field were studied repeatedly
ridors are oriented dominantly in the west/northwest direction. before and after the drilling pattern was converted from an inverted
Most of the several injector/producer short cuts are also oriented in nine-spot pattern into a vertical line drive. Approximately 37 BHI
the west/northwest direction, supporting the view that fracture cor- logs were obtained from horizontal wells, which provided valuable
ridors are responsible for the short cuts. information and paved the way toward a comprehensive under-
Flowmeter logs from vertical injector or producer wells inter- standing of fractures. The present study is the latest phase in the
secting a fracture corridor show a step profile. A comparison of the ongoing appraisal of fractures, which is aimed at approaching a
injection or production history of wells with or without a step more predictive fracture model by integrating previous findings
profile provided a means to calculate permeability enhancement by with all available BHI logs and production and seismic data.
fracture corridors. The field has more than 300 vertical wells and
nearly 20 horizontal wells, which allowed us to generate detailed Geologic Setting
fracture-permeability enhancement and fracture-corridor density The area has been very active tectonically and has witnessed sev-
maps based on injector and producer data, short cuts, mud losses, eral depositional and deformational episodes since the Early Cre-
openhole logs, and BHI logs. We also were able to build stochastic taceous (Mount et al. 1998) (Fig. 2). The main paroxysmal events
3D fracture-corridor models using corridor density from dynamic took place during the Campanian era, when Oman ophiolites were
data and orientation from BHI logs and seismic data. Fracture- obducted onto a carbonate platform (Loosvelt et al. 1996; Terken
corridor length and width were tied to fracture-permeability en- 1999). The structure was formed during this event as a low-relief
hancement using wells with both image logs and production data. fault-bend-fold over a deep-seated blind thrust fault (Fig. 4). Some
The fracture-permeability enhancement maps were verified inde- of the old Mafraq structural elements, including some northeast-
pendently by waterflood-front maps. Notwithstanding the uncer- trending faults, may have been reactivated at this compression
tainties, the fracture data were sufficiently accurate and detailed to stage. The area was subjected to repeated tectonic activity during
generate both single- and dual-porosity simulation results with Campanian and Maastrichtian times. The west/northwest- and
good field-scale history match. northwest-trending normal and wrench faults of the Oman Moun-
Introduction tain trend and related fractures must have formed during these
The field was discovered in 1968 in Oman (Fig. 1). Production is paroxysmal events. During the Tertiary period, the area was sub-
from the Shua’iba and Kharaib reservoirs of the Lower Cretaceous jected to two major tectonic activities, first in Late Eocene-
age. The Kharaib is a poorly bedded stack of repetitive shoaling Oligocene times and later toward the end of Miocene time (Fig. 2).
cycles. The Shua’iba reservoir consists of a deepening upward Some of the faults and fractures were reactivated during these
sequence. The thick-bedded massive Lower Shua’iba-B gives way Tertiary tectonic events. There are some indications on seismic
to the well-layered Lower Shua’iba-A and Upper Shua’iba units. profiles that the Eocene carbonates were exposed locally and sub-
The Kharaib and the Lower Shua’iba are separated by a horizon of jected to karstification, especially over the reactivated Cretaceous
tight argillaceous limestone (Hawar or Kharaib-K1, Fig. 2). The faults, which facilitated the mixing of fresh and salt water. The
Shua’iba reservoir is directly overlain unconformably by Pale- area underwent a regional tilting during the Late Miocene tectonic
ocene Umr er Radhuma in this field. activity. Both seismic profiles and core data suggest that some
The field consists of two low-relief eastern-A and western-B northwest-trending faults were reactivated during this latest phase.
domes. Both the Kharaib and Shua’iba units are oil-bearing in the Three unconformities separate Tertiary shales and carbonates
A and B fields. The A structure is subdivided into northern and from the underlying Shua’iba formation. Cretaceous shales are
southern fields. The southern field is located on a west/northwest preserved only at the flanks, and Shua’iba is directly overlain by
fault zone through the southern flanks of the A field (Fig. 3). A Tertiary shales and carbonates (Fig. 4). The Middle and Late Cre-
major west/northwest graben connects the A south to B. Another taceous formations are completely missing at the crest of the field
major fault zone is located on the northern flanks of the A field. (Fig. 2).
Production from the field started in 1976. Waterflooding was
started in 1986 following a series of pilot projects. The original Previous Work
An impressive amount of work already has been done on fractures.
Most of the critical questions have been answered already, and a
high degree of control over fracture distribution already has been
Copyright © 2006 Society of Petroleum Engineers
achieved. Previous studies and significant results of the fractures
This paper (SPE 93312) was first presented at the 2005 SPE Middle East Oil and Gas Show are summarized below before we discuss our findings.
and Conference, Manama, Bahrain, 12–15 March, and revised for publication. Original
manuscript received for review 8 January 2005. Revised manuscript received 9 February
A detailed structural, sedimentological, and petrographic analy-
2006. Paper peer approved 1 March 2006. sis by Petroleum Development Oman (PDO) on the cores and BHI

June 2006 SPE Reservoir Evaluation & Engineering 227

Fig. 1—Location map of the North Oman fields.

logs from a vertical well showed that faulting and fracturing took
place mainly during Late Cretaceous emplacement of thrust sheets
in the Oman Mountains. The vast majority of natural fractures are
cemented and therefore represent permeability baffles.
Arnott and Van Wunnik (1996) presented the injector/producer
infill patterns implemented to make optimum use of the fractures.
They outlined a strategy to manage variable risk of drilling close
to previously unidentified faults. These authors noted that a num-
ber of faults are often characterized by narrow but intense small
fracture zones and have a clear association with early water break-
through from nearby injectors. The fracture-density variation along
the well has similar patterns for both formations, but Shua’iba-A is
more fractured than Kharaib.
Lekhwair fractures were studied previously by two PDO ge-
ologists (Everts and Leinster) to characterize downflank quality in
the A and B areas and to quantify fault and fracture geometry from
core and BHI logs. They noted that deterioration of reservoir qual-
ity within the down-flank areas may well be caused by pressure
solution affecting the reservoir after hydrocarbon migration. The
occurrence of a tightly cemented matrix around the damage zones
suggests that faults will have limited connectivity with the matrix
and may compartmentalize the reservoir. Anomalous fluid path- Fig. 2—North Oman stratigraphy (Al-Busaidi 1997).
ways are more likely to be localized (karst pipes) rather than a
fieldwide open network of fractures. Close inspection reveals no
consistent relationship between the occurrence of (BHI-identified) jointing. Observations that support this conclusion include a high
fracture clusters and the presence of flushed zones. degree of fracture clustering and an occurrence of fracture clusters
Al-Busaidi (1997) discussed the use of BHI logs and reached near faults. Fracture orientations from BHI logs reveal that the
the opposite conclusions. He saw fractures as the primary cause of dominant fracture strike is parallel to the west/northwest Oman
early water breakthrough, based on integration of BHI logs and Mountain trend with minor fracturing in the northeast and north-
production logs. Densely fractured zones in most cases correspond west directions. Fracturing seems to be controlled by stratigraphy
to faults, and the number of open fractures in a well is correlated to some extent. In general, all authors note that the Kharaib is far
with water-cut percent. less fractured than the Lower Shua’iba reservoir unit. Existence of
A subsequent in-house study focused on the regional tectonic multiple phases of fracturing is evidenced by different episodes of
elements and attempted to identify fractures associated with Ma- calcite cementation detected by petrographic studies and the cross-
fraq and Oman Mountain trends on the basis of fracture orienta- cutting relationship of stylolites and fractures.
tion. The field is affected by three main fracture orientations, One point of divergence is the degree of cementation that frac-
which are related to regional fault trends. The dominant fracture tures have undergone. Some PDO geologists regarded all open
orientation is west/northwest parallel to faults of the Oman Moun- fractures as drilling-induced fractures and were very skeptical
tain trend. about the existence of fluid-conductive open fractures. Arnott and
There is a general consensus that the majority of fractures are Van Wunnik (1996) and Al-Busaidi (1997) had the opposite view
fault-related and are superimposed on a widely spaced background and held faults and fractures responsible for the early water break-

228 June 2006 SPE Reservoir Evaluation & Engineering

Fig. 4—North-to-south seismic profile through the A-sector of
the field, showing the deep-seated thrust fault, normal faulted
flanks, and base tertiary unconformity (see Fig. 3 for location).

graphic evaluation. Seismic data were used to determine whether
it is possible to predict fractured zones from seismic data.
We evaluated fracture flow potential in the next phase of the
work, cross-validating our findings by indirect fracture flow indi-
cators such as mud-loss occurrences, water fingering, distribution
of fracture injectors, and wells with high water cut. Open fracture
clusters, which coincided with mud losses and water breakthrough,
are regarded as high-confidence fluid-conductive fracture corri-
dors. A set of low-confidence fracture corridors was identified
from wells with mud losses, high water cut, fracture injectors, and
early water breakthrough. Orientation of these low-confidence
fracture corridors is estimated by the orientation of nearly-high-
confidence fracture corridors.
After fracture analysis, we made an attempt to generate a map
model of fracture corridors. The procedure was started by gener-
ating maps of high-confidence fracture corridors and continued by
mapping low-confidence corridors. Fracture clusters detected in
BHI logs are regarded as high-confidence corridors. Low-
Fig. 3—Base map of the field A-North and A-South sec-
tors showing seismic faults, well locations, and horizontal-
confidence corridors correspond to mud losses, high-water-cut
well trajectories. wells, fracture injectors, and water fingering. The composite maps
were used to interpolate between fracture corridors to generate a
possible fracture-corridor map with anticipated position, orienta-
through and short cuts observed in many wells. The drilling pattern tion, length, and width. The interpolated fracture fairway map was
was changed on the basis of the hypothesis that faults and fractures subdivided into sectors, with each sector having nearly uniform
are fluid-conductive. fracture characteristics. Tables were prepared to summarize frac-
ture-corridor statistics for each sector to provide fracture data for
Data Sources and Procedure future reservoir simulation.
The main source of fracture information is BHI logs from more In the final phase of analysis, critical questions were
than 35 horizontal wells and cores from one horizontal well and addressed such as the timing of fractures, reactivation, and under-
the vertical well. Stratigraphic information (which was accumu- lying causes for the observed distribution of cemented and open
lated over the years from boreholes) and structural features (such fracture corridors.
as faults from 3D seismic data) constitute the framework for frac- Five main fracture types are recognized in the Shua’iba and
ture evaluation. The fracture data from BHI logs were analyzed in Kharaib reservoirs of the field:
conjunction with dynamic flow data such as mud-loss occurrences, • Dispersed background joints.
gross production and injection rates, water cut, and water short • Clustered fractures.
cuts. Most of the horizontal wells are in the Lower Shua’iba A and ‫ ؠ‬Joint swarms.
B units. Only five wells are in Kharaib or have lower laterals in ‫ ؠ‬Fracture corridors.
Kharaib. Several wells intersect some section of the Upper • Disrupted zones.
Shuaiba, and only one well is in the Upper Shua’iba itself. The • Megafractures.
horizontal wells provide a good coverage of the A-north sector. • Faults.
Most A-north wells are oriented northeast and, hence, may under- The dispersed background fractures are typically observed as very
sample northeast-trending fractures. widely spaced, layer-bound small joints. The dispersed fractures
The work started by identifying and distinguishing layer-bound are mostly mineralized, although occasionally a few nonmineral-
joints from fault-related hybrid fractures and fracture clusters. Al- ized joints have been observed in BHI logs.
though some coring-induced fractures were observed in cores, no Fracture clustering can be interpreted in two different ways
drilling-induced fractures are present in the image logs from hori- (Figs. 5a and 5b). A cluster is either a joint swarm or a fracture
zontal wells. We first examined fractures that are identified on BHI corridor. Joint swarms are layer-bound fractures. A joint swarm is
logs within a stratigraphic framework, searching for a correlation observed in BHI logs when a horizontal well traverses a thin-
between fracture density and layer thickness in particular. Exam- bedded and highly fractured unit. Fracture corridors are clusters of
ining fractures in relation to structural aspects followed the strati- fractures that are confined to subvertical narrow tabular bands


fractures are few in number, but they are believed to play a sig-
nificant role in shaping the reservoir flow dynamics because these
fractures have large apertures (ranging from 0.2 to 0.5 mm).
A disrupted zone is characterized by one or more of the fol-
lowing features: (1) bedding disturbance, (2) fragmentation, and
(3) cementation (Fig. 5c). Disrupted zones are often small faults,
fault splay, or deformation bands, but these could be related to
other causes such as collapsing karstic cavities. Megafractures
with large apertures characteristically occur within fracture corri-
dors (Fig. 5d), but a few isolated ones were also observed, which
may be occasionally large open joints. Megafractures may have
cemented walls.

Fracture Corridors
A previous fracture study by Ozkaya et al. (2003) shows that
fractures in most Oman carbonate fields consist of dispersed layer-
bound fractures and fracture corridors (Fig. 6).
The previous study by Ozkaya et al. (2003) also suggested that
only fracture corridors play a significant role in reservoir flow
dynamics. Dispersed background fractures have limited flow po-
tential and may be treated as a matrix property. The present study
places emphasis on the detection of fluid-conductive fracture cor-
ridors and quantifies their attributes, such as permeability enhance-
ment and density distribution. Therefore, identification of fluid-
conductive fracture corridors is one of the main objectives of this
study. Fluid-conductive fracture corridors are identified on BHI
logs from dominance of electrically conductive fractures. Fracture
corridors often can be identified in BHI logs from the high degree
of clustering and the association of large and small fractures. In
some cases, it is not possible to determine whether a fracture
cluster is a corridor or a highly fractured layer, and it is necessary
to probe further into the nature of the cluster by correlating the
cluster with bedding thickness and faults (Fig. 7a). Fracture clus-
ters that can be correlated between lower and upper laterals or
nearby horizontal wells are interpreted as fracture corridors regard-
less of faulting (Fig. 7b).

Fig. 5—BHI examples. (a) and (b) show fracture clusters, which
probably represent fracture corridors. (c) is a disrupted fault
zone, and (d) is a large conductive fracture (megafracture).

(Figs. 5a and 5b). Fracture corridors are often, but not always,
associated with faults. Tight curvatures, flexures, or deformation
bands may also generate fracture corridors. Layer-bound mega-
fractures often result from alignment of layer-bound fractures Fig. 6—Schematic fracture model of the Shua’iba and Kharaib
(Hedgeson and Aydin 1991; Cooke and Underwood 2001). Mega- reservoir units.


Fig. 7—Fracture-corridor identification by fault association (a)
and by correlation of multilaterals (b). Fig. 8—Stick plot map of high-confidence fluid-conductive and
cemented fracture corridors.

The association of fracture corridors with mud losses is a strong
indication that the fracture corridor is fluid conductive. If mud is sharp drop in oil saturation with no corresponding change in po-
well balanced, mud loss is minimal and may be overlooked. A rosity often indicates either a fracture corridor or water encroach-
cemented cluster has no (or very few) open fractures and does not ment through the matrix. A comparison with BHI logs and pattern
correspond to any mud loss or indication of fracture flow by way recognition may allow identifying corridors in the absence of BHI
of high water cut. logs (Fig. 9). Openhole logs from horizontal wells also can be used
Once a fracture corridor is identified on the BHI logs, its to differentiate water fingering through fracture corridors and
attributes must be quantified. The critical attributes of corri- highly permeable matrix streaks. Broad zones with high water
dors include location; width; the number of open, cemented, saturation (the wide bands in Fig. 9) represent water encroach-
and megafractures; corrected fracture spacing; and orientation. ment through permeable layers, whereas sharp spikes represent
These attributes were quantified for all fluid-conductive and ce- fracture corridors.
mented corridors. Flowmeter Logs. Flowmeter logs are extremely useful in the
Our study on the fractures from BHI logs revealed that open identification of fracture corridors and the determination of their
fracture corridors are confined to the near crest of the A-north flow potential. Fracture corridors often cause highly prominent
sector. Fracture corridors on the flanks are cemented. A possible steps in flow profiles. Fracture-permeability enhancement can be
explanation is that at least one stage of fracturing took place after calculated from the magnitude of the step. Step size also differen-
oil emplacement. Fractures within the oil leg remained open, while tiates between fracture corridors and high-permeability streaks,
those within the water leg were sealed by carbonate cement which are often located near the top of Lower Shua’iba-B. The
(Fig. 8). The increasing degree of fracture cementation downflank ratio of rate of flow from step profiles to total flow outside step
is also accompanied by the deterioration of reservoir quality, profiles is an indication of the transmissivity ratio of the feature
which is attributed to pressure solution affecting the reservoir after causing the step profile to the total matrix reservoir transmissivity.
hydrocarbon migration (Terken 1999). Transmissivity of fracture corridors ranges between two and seven
times the total matrix transmissivity (permeability times reservoir
Identification of Fluid-Conductive Fracture Corridors in the thickness, kh). Transmissivity of high-permeability streaks ranges
Absence of Image Logs. Only a limited number of BHI logs are between 0.25 and 1.25 times the total reservoir matrix transmis-
available. The horizontal wells with image logs are mostly on the sivity. High-permeability streaks were identified in 4 of the 32
flanks, leaving a large gap on the crestal area with no image logs flowmeter logs. A total of 10 (out of 32) flowmeter logs show
(Fig. 8). Therefore, it is necessary to resort to other indicators of conspicuous step profiles that are interpreted as fracture corridors.
fracture corridors, which include openhole logs, flowmeter logs, BHI and flowmeter logs form a powerful combination to quan-
and well performance histories from producer and injector wells. tify fracture corridors and tie in the flow potential to fracture
Openhole Logs. In horizontal wells, openhole log plots often density and corridor width. Unfortunately, only a small number of
help identify and locate fluid-conductive fracture corridors. A vertical injector wells and vertical producer wells have flowmeter


Fig. 10—A bubble map of initial PI values from approximately
20 wells.

Fig. 9—Porosity and water saturation from openhole logs and PI measurements are available for approximately the first 70 wells,
fracture density from image logs in a horizontal well. and only about 20 of those are from the A-North sector (Fig. 10).
The limited number of wells and variable timing reduce the value
logs. There are no horizontal wells with both flowmeter and of PI for identification of fracture corridors. Nevertheless, the
BHI logs. high PI values are within the faulted southern flank of the A-
Mud Losses. Mud losses are mostly indicative of fluid- north structure.
conductive fractures and faults, but karstic features and high mud Total horizontal transmissivity (kh) is also highly indicative of
pressure and hydrofracturing may also cause mud losses. Mud fracture flow. Total kh values are calculated for approximately the
losses are used as a backup and as cross-validation data in this first 70 wells, with only about 20 being from the A-north sector of
study, but not for the location of fracture corridors. Mud losses that the field. The high values seem to be concentrated within the
are coincident with fracture corridors are the prime indicator of southern faulted sector, with a few high values at the crest
fluid-conductive fracture corridors on BHI logs. (Fig. 11).
Productivity Index and Total Well Permeability. Productivity Well Performance Histories. Well performance histories are
index (PI), especially initial PI, is a good indicator of fracture flow. perhaps the most important source of information for identification
of fluid-conductive fracture corridors. High gross sustained pro-
duction rates may be an indication of fracture flow. Average sus-
tained gross is calculated from well performance history after ex-
cluding the initial decline and eliminating all irregularities, such as
shut-in periods. Map distribution provides an insight into the dis-
tribution of fluid-conductive fracture corridors. Wells show high
rates in the southern and northern faulted flanks, while the wells in
the middle crestal sector of the field are mostly low-rate matrix
producers (Fig. 12).
High injection rates and injection rate divided by tubinghead-
pressure values also may be indicative of fracture flow. Injection
rates are calculated from well performance history. Only the initial
injection phase is taken into consideration before injection rates
start to decline. Average injection rates divided by tubinghead-
pressure values are used when there are changes in tubinghead
pressure during injection. Injection-rate bubble plots indepen-
dently confirm indications by producer wells that most fracturing
is concentrated on the southern and northern faulted sectors
(Fig. 13).
Wells Intersecting Fracture Corridors. Matrix and fracture
producers are differentiated on the basis of sustained gross rates.
The threshold value is estimated to be 150 m3/d by averaging the
gross rates of wells with gradual flowmeter profiles. This threshold
value is the total combined gross rate for the Lower Shua’iba and
Kharaib reservoirs. Wells with gross rates higher than 150 m3/d are
regarded as fracture producers. Such wells are assumed to intersect
Fig. 11—Bubble map of kh from approximately 20 wells in the fluid-conductive fracture corridors. The threshold value is vali-
North sector of the field. dated by flowmeter logs. A well with less than 150 m3/d produc-


Fig. 12—Sustained gross rates. Fig. 13—Normalized injection rates.

tion has no conspicuous step profiles that can be interpreted as
fracture corridors. Wells Fractured by Injection. The injection history of approxi-
Matrix and fracture injectors are separated on the basis of initial mately 10 injector wells suggests hydraulic fracturing by excess
injection rates divided by tubinghead pressures. The matrix/ injection pressures. In these wells, injection rate jumps up sud-
fracture threshold value is estimated as 0.08 m3/d/kPa on the basis denly, with no corresponding change in injection pressure. The
of fracture and matrix injectors. The threshold rate is estimated as injection rate of some of these wells may reach injectors that
the average injection rate of wells with gradual flowmeter profiles. intersect natural fractures, probably because injection-induced
Step profiles identify fracture injectors. fractures link to the natural fracture system. These were regarded
Because a large number of vertical producer and injector wells as matrix injectors.
are available in the field, the field-scale distribution of fluid- Pressure-Transient Analysis. Well tests are very useful for
conductive fracture corridors is based mainly on these data quantifying fracture-corridor length and permeability; unfortu-
(Fig. 14). From the map distribution of fracture producer or injec- nately, no well tests were available when this study was conducted.
tor wells, a clear concentration in the southern and northern faulted
sectors is seen. In addition, the map distribution also reveals the Direction of Fracture Corridors. Producer or injector well per-
alignment of fracture wells along some open or hidden fault zones, formance or other indirect fracture-flow indicators do not provide
such as the north/northwest fault on the western flank and close to information on the orientation of fracture corridors. A strike and
the crest. dip must be assigned to each fracture corridor. The following
sources of information were used to assign direction to fracture
corridors from production data.
BHI Logs. The corridors from image logs have been identified
and mapped, and the dominant orientation of fracture corridors and
secondary sets have been determined (Fig. 8). Most of the fracture
corridors on BHI logs are oriented in a northwest direction. There
are two subordinate sets with north/northwest and northeast direc-
tions. This information is used to assign an orientation to corridors
from production data in two ways:
• Statistically, the percentage of northwest, north/ northwest,
and northeast corridors from production data are made equal to
the relative abundance of corridors in these directions from
BHI logs.
• Corridors with unknown orientation are assigned an
orientation similar to nearby corridors from BHI logs with
known orientation.
Seismic Faults. Corridors in BHI logs and injector/producer
short cuts are found to be parallel to seismic faults. This observa-
tion provides an additional means to assign orientation to corridors
from production data. Corridors near seismic faults are assigned
orientation parallel to the seismic fault.
Injector/Producer Fracture Short Cuts. An extensive investi-
gation was performed to locate injector/producer short cuts by
comparing the well performance histories of injectors and nearby
producer wells. This information is used to demonstrate the
existence and length of large-scale corridors (fairways) and
Fig. 14—Fracture producers: wells intersecting fracture corridors. their orientation.


Dip Angle of Fracture Corridors. For horizontal wells with im- ment. Grid data for fracture-permeability enhancement have been
age logs, the corridor dip angle and dip azimuth are taken from the generated for the three main directions northwest, north/northwest,
BHI logs. If a vertical well is on a corridor or fairway intersected and northeast. These are the essential fracture data to be incorpo-
by horizontal wells with BHI logs, it is assigned the same dip rated in a single-porosity simulation model.
angle. If a vertical well is away from known corridors, the average
corridor dip is assigned to it; the average corridor dip is calculated Cross Validation
from BHI logs. An attempt has been made to validate fracture results by using
additional data sources. Flood-front maps are used to validate the
Length of Fracture Corridors. Length of fracture corridors can map distribution of fracture corridors from production data. Some
be estimated by correlating fracture corridors in adjacent horizon- flowmeter logs from producing wells and a few additional injector
tal wells. Similarly, corridor height can be estimated from dual wells are used to confirm fracture and matrix wells identified from
laterals (Fig. 7). Unfortunately, only a limited number of wells other sources. Injector wells and one crestal well with an image log
have dual laterals or are close enough to allow correlation of are used to validate the ratio of corridors in the Lower Shua’iba
corridors. We have estimated average corridor lengths as 150 m and Kharaib. Water fingering from openhole (OH) logs is used
from the width/length ratio of the corridors from two nearby to validate flood-front maps from BSW (bulk solids and
wells and the average width of corridors in other wells. We have water—water cut). Reported faults from openhole logs with mud
adopted a 3-to-1 length-to-height ratio based on the correlation losses are used to validate fluid-conductive corridors from hori-
of corridors between lower and upper laterals in two wells with zontal wells.
dual laterals.
Individual corridors from production data are assigned lengths Flood-Front Movement. Flood-front movement is one of the key
on the basis of the permeability enhancement they cause. The indicators of fracture flow. Therefore, water-movement maps were
average permeability enhancement is assigned to the average correlated with the fracture-corridor density maps. Flood-front
corridor length of 150 m. The corridor length is zero for ma- maps were prepared to show wells with more than 60% BSW in
trix producers. 1995 and 2000 (Fig. 17). There is a major convergence between
high water cut and fracturing. Most of the water comes through
Fracture Corridors Encountered in Wells. Fracture corridors fractures except on the eastern flank, where some matrix water
are generated from production and image log data (Fig. 14) after encroachment is observed. This encroachment is caused by the
orientation, and length is assigned to each corridor. As noted ear- Lower Kharaib bringing in water because of its proximity to free
lier, orientation is based on BHI logs, adjacent seismic faults, and water contact. Although not indicated on these maps, high water
short cuts. Length is based on permeability enhancement. Fracture- cut has little correlation with the location of injector wells.
corridor length must be regarded as an index, rather than the actual Previous examination of injector flowmeter logs revealed the
length, because no reliable length estimate is available. existence of a high-permeability streak at the top of Lower
Shua’iba B. In combination with fracture corridors, such
Fracture Fairways and Short Cuts. The average length of frac- high-permeability streaks may facilitate water encroachment and
ture short cuts is more than 1 km, which suggests the existence of overriding, causing poor sweep efficiency. To answer this
interconnected fracture corridors. Strong alignment of corridors, question, water-encroachment maps were prepared from produc-
especially in the vicinity of seismic faults, supports the idea that tion data.
such interconnected clusters of fracture corridors do exist. Water fingering was also mapped from vertical-well openhole-
log saturation and porosity data for the Shua’iba A and B and the
Fracture-Permeability Enhancement Maps Kharaib-5, respectively. Water-encroachment data are taken as the
Fracture-permeability enhancements from producer and injectors difference between average observed and expected hydrocarbon
wells are presented as bubble plots in Fig. 15 and as a contour map saturation in the three reservoir intervals Lower Shua’iba A and B
in Fig. 16, with fracture corridors superimposed to show the mag- and Kharaib-5. Kharaib-5 is the fifth subunit of the Kharaib res-
nitude as well as the direction of fracture-permeability enhance- ervoir from the top and is the main Kharaib subunit.

Fig. 15—Fracture-permeability index. Fig. 16—Fracture-permeability index contour map.


Fig. 17—Flood-front movement map. Fig. 18—A snapshot of water encroachment in Lower Shua’iba-A.

One drawback of this method is that the openhole-log satura- The water-saturation bubble map from openhole logs of verti-
tion is obtained immediately after drilling and reflects only the cal wells shows that most of the water fingering in Lower
situation at the drillsite at the time of drilling. Many of the Shua’iba-B starts at faults and fracture corridors (Fig. 20). This
study wells are, on average, 20 to 30 years old. Therefore, the map also can be interpreted as an indication of the waterflooding
results cannot be regarded as a reflection of present-day actual mechanism. Water moves through faults and fracture corridors into
water encroachment. high-permeability streaks within the Lower Shuiba-B unit first.
In all three reservoir units, water fingering starts in the frac- The type of fracturing also may encourage early water break-
tured northern and southern sectors, confirming the findings from through in the B unit, rather than the A unit. Although A is thinly
production data. Comparison of the maps shows that water finger- bedded and far more fractured than B, the B unit has considerably
ing is most advanced in Lower Shua’iba-B (Figs. 18 and 19). more megafractures (large conductive fractures).
Porosity and saturation logs from horizontal wells confirm this
observation (Fig. 9). Of 21 horizontal wells, only 4 show slight Validation of Corridor Concept With Flowmeter Logs. This
flushing in Lower Shua’iba-A (only 2 in the Kharaib). The re- study is fundamentally based on the assumption that fracture cor-
maining 15 wells show partial or complete flushing of the ridors are the main fracture-flow conduits. Dispersed fractures
Shua’iba-B unit. have little flow potential either because of cementation or because

Fig. 19—A snapshot of water encroachment in Lower Fig. 20—Rise in water saturation from openhole logs acquired
Shua’iba-B. at different times in the Lower Shua’iba-B.


of their small size and lack of connectivity. Flowmeter logs pro- suggests that the threshold value classifies most of the wells
vide a means of validating this assumption. When a borehole in- correctly. The error margin based on flowmeter logs is estimated
tersects a fracture corridor, one or a few megafractures within the to be 10%.
corridor are expected to act as a permeability spike and cause a
step profile in flowmeter logs. A pervasive fracture system is Fracture-Corridor Length. Average corridor length estimation is
expected to generate a smooth flow profile, even though total flow based on the length/width correlation from the BHI log and the
rate is higher than matrix rates. The flowmeter logs from fracture only two adjacent wells to calibrate the results. Corridor length is
injectors or producers are always marked by a step profile, sug- also estimated indirectly by matching corridor density from BHI
gesting corridors rather than pervasive layer-bound fracturing. logs with corridor density from production data. In both cases,
there is a large degree of uncertainty. The uncertainty in length is
Validation of Corridor Density and Length From BHI Logs. BHI not critical if only fracture-permeability enhancement grid data are
logs provide a reliable means of measuring the scan-line density of used to upscale and import fracture data into simulators. It be-
fracture corridors at different sectors of the field. These values are comes critical, however, if fracture data are generated by upscaling
used to validate the corridor-density calculations from production stochastic models.
data as follows. Four crestal wells are used to estimate corridor
scan-line density within the highly fractured and faulted southern Corridor Density. Corridor density is calculated by counting the
sector, the crestal area, and the northern faulted sectors, respec- number of fracture producer or injector wells within a circular
tively. These results of the corridor scan-line density from BHI window. Corridor length, height, and shape affect density-grid
logs and production data are compared. There is good agreement calculations to some degree. Although it is difficult to quantify the
between the two results. For example, the north/south crestal wells uncertainty, we found that changing the window size affects only
yield an average corridor spacing of 161 m for northwest corridors. the density spread, not the actual values. Fracture height is a criti-
Production data yield 125- to 167-m spacing in the highly frac- cal factor, and we have little control over the height of corridors
tured and faulted sector in which the wells were drilled. except in the assumption that corridors either cut through both
reservoir units or are confined to one unit only.
Uncertainties
Defining uncertainties is an important step in fracture-data prepa- Fracture Communication Between the Kharaib and Lower
ration for simulation. We have made an attempt to identify the Shua’iba. There are no data to evaluate the degree of communica-
main uncertainties and quantify them whenever possible. These are tion between the Lower Shua’iba and Kharaib reservoirs through
explained below. the low-permeability Hawar muddy carbonates. Some fracture cor-
ridors are identified in both Kharaib and Lower Shua’iba in wells
Are Fractures Responsible for Permeability Enhancement? The with dual laterals. However, this does not provide any information
fundamental question is the connection between fractures and per- on pressure communication. The fracture corridors may consist
meability enhancement; no wells have both image logs and flow- only of a few shear fractures within the Hawar unit with no pres-
meter logs to definitely match and identify fluid-conductive cor- sure communication.
ridors. Mud losses are highly supportive but do not constitute
definite proof because mud losses may also be associated with
karstic dissolution features, or hydraulic fractures. The first main Relative Significance of Corridors and Dispersed Background
observation is that BSW is high when gross production rate is Fractures. The present work is based on the assumption that frac-
also high. The second observation is that the high BSW and ture corridors are the main flow conduits. Flowmeter logs have
high production and injection-rate wells are clustered in the south- been used to validate the corridor concept (see the previous sec-
ern and northern highly faulted sectors of the field. These two tion), but step profiles do not constitute solid proof of that concept.
observations suggest that fractures may be the cause of permeabil- In some wells, wide zones of fractures, which merge to form a
ity enhancement. pervasive fracture system, surround corridors. This is even more
If fractures are responsible for permeability enhancement, per- conspicuous in the thin-bedded Lower Shua’iba-A. The relative
meability enhancement should correlate with corridor width and flow potential of narrow corridors with megafractures in compari-
open fracture spacing. Permeability enhancement of fracture cor- son to wide bands of dispersed joints is highly uncertain at this
ridors, which are intersected by both vertical wells and horizontal stage. The fundamental observation, however, remains unchanged.
wells with BHI logs, makes it possible to calculate the correlation Both corridors and dispersed fractures are more abundant in the
coefficients. The correlation coefficients between permeability en- highly faulted northern and southern sectors of the field.
hancement and corridor width and open fracture spacing are
0.31 and 0.42, respectively. These correlation coefficients are Fracture-Corridor Distribution Between Lower Shua’iba and
low and fail to remove the uncertainty surrounding the funda- Kharaib. One important question is the distribution of fracture
mental assumption that fractures are responsible for permeabil- corridors between the Lower Shua’iba and Kharaib reservoirs.
ity enhancement. This question is related to (i) rock-mechanical aspects of reservoir
units, (ii) the mechanism of corridor generation, and (iii) cemen-
Threshold Values for Fracture/Matrix Producer and Injec- tation. The only well with dual laterals in the Lower Shua’iba and
tors. The bulk of the fracture-corridor information is based on the Kharaib shows that the Lower Shua’iba is more fractured than the
performance history of the producer and injector wells. The matrix Kharaib and has more open fractures and corridors (Fig. 7). The
and fracture producers and injectors are classified on the basis of frequency of corridors in Lower Shua’iba or Kharaib is highly
arbitrary threshold values. The frequency distribution of gross pro- variable depending on the source of information.
duction and injection rates is log-normal, with no obvious differ- Injector flowmeter logs and well image logs suggest a 3-to-1
entiation of fracture and matrix producers. The threshold values ratio for the Lower Shua’iba and Kharaib corridors. However,
are close to modal values in both cases, which implies a high fracture producer and injector data suggest a 1-to-1 ratio. Obvi-
degree of sensitivity. For example, a change from 150 to 175 ously, BHI and flowmeter data must be regarded as more reliable
would eliminate 40 wells as fracture producers. In the case of than production data in this respect because in many wells, it is
injector wells, the sensitivity is less. A change in the threshold not possible to determine conclusively which unit intersects a frac-
value from 0.08 to 0.1 would eliminate only 10 wells as fracture ture corridor.
injectors. Some matrix wells intersecting high-K matrix streaks
may have production rates higher than the 150 m3/d threshold Clustering of Fracture Corridors. It appears that corridors are
value. Some of the wells, which are interpreted as fracture pro- more abundant in the vicinity of major faults, but no analysis has
ducers or injectors at the crest of the field, may actually intersect been done to ascertain the affinity of corridors to major faults.
only high-K matrix streaks. The general map pattern, however, Plotting the distance of corridors from the nearest fault as a fre-


quency plot and measuring deviation from random distribution seismic faults and production data allows us to generate the
could provide a measure of the degree of corridor clustering. following for the three main fracture directions:
• Horizontal fracture-permeability enhancement map for both
the Lower Shua’iba and Kharaib reservoirs.
Stochastic Modeling • Fracture-corridor density maps and stochastic fracture
A viable procedure to incorporate fracture-corridor data into a models. Each stochastic and observed corridor is assigned
simulator is to generate stochastic corridor models, which can be necessary attributes, including permeability enhancement
used to extract fracture parameters for dual-porosity simulation for upscaling.
such as matrix block size and effective fracture permeability and 5. The horizontal fracture-permeability enhancement maps may
porosity. For this purpose, discrete stochastic models of fracture be used to generate vertical-fracture enhanced permeability
corridors are generated using the corridor density-grid data for maps, which can be imported directly into the reservoir simulator.
undifferentiated corridors. In these models, corridor distribution is 6. Alternatively, the stochastic fracture models can be used to gen-
assumed to follow the Poisson distribution. Corridor length is as- erate upscale fracture data into the simulator. In this case,
sumed to have exponential distribution with an average of 150 m. it is necessary to perform single-well simulation runs to con-
Corridors are simulated as three distinct (northwest, north/ vert the fracture-permeability enhancement into fracture-
northwest, and northeast) sets. The corridor orientation has a uni- corridor transmisisvity.
form distribution with 10° range around the average strike of these The main uncertainties are:
three sets. 1. Threshold value to differentiate fracture producer or injector
Fig. 21 shows a stochastic realization. The actual corridors are wells from matrix producer/injectors.
superimposed on stochastic corridors. An equal number of sto- 2. Fracture-corridor length.
chastic corridors are removed to maintain the observed corridor 3. Fracture communication between the Kharaib and Lower
density. The location of stochastic corridors ignores wells that did Shua’iba reservoirs.
not actually intersect corridors (matrix producers). Some of the 4. Corridor distribution between Lower Shua’iba and Kharaib.
corridors at the margins are removed because these are outside the 5. Degree of clustering of fracture corridors.
area of interest and generated by extrapolation. The width of the Notwithstanding the uncertainties, the fracture data are sufficiently
corridors in plan view reflects the dip angle and height. Each accurate and detailed for reservoir-simulation purposes. The
corridor in these stochastic models is assigned a permeability en- following additional data acquisition may be necessary to reduce
hancement value based on its length and the relationship between the uncertainties:
length and permeability enhancement for the actual corridors to 1. Tests to establish pressure communication between Lower
allow upscaling. Shua’iba and Kharaib.
2. Measurements to evaluate present-day oil saturation using sur-
veillance data.
Conclusions 3. Additional flowmeter logs for better control of the frac-
The main conclusions are as follows: ture corridors.
1. BHI and seismic data show that the main fracture direction is 4. Map distribution of bottomwater and injector water produced.
northwest, with subordinate sets in the north/northwest and 5. Assessment of the imbibition characteristics of the matrix/
northeast directions. fracture system.
2. Fault-related fracture corridors are the main fracture-flow con- 6. Performance of several well tests to quantify fracture-corridor
duits. Dispersed background joints have little flow potential length and transmissvity.
because of cementation, lack of connectivity, or small size.
Acknowledgments
3. Fracture corridors are fluid-conductive only within the oil leg
but are cemented at the flanks within the water leg. We would like to thank Petroleum Development Oman and the
4. Integration of available BHI, flowmeter, and openhole logs with Oman Ministry of Oil for permission to present this work. A.J.
Everts and R.C. Leinster conducted a study entitled “Fractured
Carbonate Reservoir Modeling—Lekhwair Field” in 1997; their
work is available as an internal PDO report.
References
Al-Busaidi, R. 1997. The Use of Borehole Imaging Logs to Optimize
Horizontal Well Completions in Fractured Water-Flooded Carbonate
Reservoirs. GeoArabia 2 (1): 19–32.
Arnott, S. and Van Wunnik, J.N.M. 1996. Targeting Infill Wells in the
Densely Fractured Lekhwair Field, Oman. GeoArabia 1 (3): 405–416.
Cooke, M.L. and Underwood, C.A. 2001. Fracture Termination and Step-
Over at Bedding Interfaces Due to Frictional Slip and Interface Open-
ing. J. of Structural Geology 23 (2–3): 223–238.
Hedgeson, D.E. and Aydin, A. 1991. Characteristics of Joint Propagation
Across Layer Interfaces in Sedimentary Rocks. J. of Structural Geol-
ogy 13 (8): 897–911.
Loosvelt, R.J.H., Bell, A., and Terken, J.J.M. 1996. The Tectonic Evolu-
tion of Interior Oman. GeoArabia 1 (1): 29–51.
Mount, Van S., Crawford, R., I., S., and Bergman, S.C. 1998. Regional
Structural Style of the Central and Southern Oman Mountains: Jebel
Akhdar, Saih Hatat and the Northern Ghaba Basin. GeoArabia 3 (4):
475–490.
Ozkaya, S.I., Kolkman, W., and Amthor, J. 2003. Mechanical Layer-
Dependent Fracture Characteristics From Fracture Density vs. TVD
Cross Plots. Examples From Horizontal Wells in Carbonate Reservoirs,
North Oman. Paper presented at the AAPG Intl. Exhibition and Con-
ference, Barcelona, Spain, 21–24 September.
Terken, J.M.J. 1999. The Natih Petroleum System of North Oman.
Fig. 21—Stochastic fracture-corridor model. GeoArabia 4 (2): 157–180.


ture studies in various carbonate and clastics fields in the
SI Metric Conversion Factors Middle East. Ozkaya is the author of several publications on
computer applications in fracture and structural analysis. He
bar × 1.0* E+05 ‫ ס‬Pa
holds an MS degree in computer science and a PhD degree in
bbl × 1.589 873 E–01 ‫ ס‬m3 structural geology from the U. of Missouri. Pascal D. Richard is a
ft × 3.048* E–01 ‫ ס‬m senior structural geologist at the Study Centre of Petroleum
mile × 1.609 344* E+00 ‫ ס‬km Development Oman (PDO) and Carbonate Technology Co-
ordinator for PDO in Muscat, Oman. Previously, he joined Shell
*Conversion factor is exact. in 1991 and spent 4 years in the Research Structural Geology
Dept. working on the modeling of structural styles, fault growth,
and hydrocarbon systems. After working for 3 years with PDO
as a structural geology consultant and seismic interpreter, he
Sait I. Ozkaya is a geological consultant with Baker Atlas Geo- worked for 5 years on Shell E&P’s Research Carbonate Devel-
science in Manama, Bahrain. He worked as a professor of ge- opment Team, focusing on the characterization and modeling
ology at several universities and as an exploration geologist for of fractured reservoirs. Richard is now implementing fracture
Chevron before joining Baker Atlas Geoscience in 1996. He technology at PDO and has a role in coaching and knowl-
specializes in integrated fractured reservoir characterization edge transfer. He holds a PhD degree in structural geology
using static and dynamic data and has conducted many frac- from the U. of Rennes, France.


7 spe-93312-pa-p fractured reservoir characterization using dynamic data in a carbonate field

More Related Content

What's hot (20)

Similar to 7 spe-93312-pa-p fractured reservoir characterization using dynamic data in a carbonate field (20)

Recently uploaded (20)

7 spe-93312-pa-p fractured reservoir characterization using dynamic data in a carbonate field