Faults and fractures were interpreted using attributes that were extracted from a 3-D seismic data set recorded over a Lower Cretaceous Thamama oil field in offshore Abu Dhabi, United Arab Emirates. The Thamama reservoir has good matrix porosity (frequently exceeding 20%), but poor permeability (averaging 15 mD). Because of the low permeability, faults and fractures play an important role in fluid movement in the reservoir. The combination of the similarity and dip attributes gave clear images of small-displacement fault geometry, and the orientation of subseismic faults and fractures. The study better defined faults and fractures and improved geomechanical interpretations, thus reducing the uncertainty in the preferred fluid-flow direction. Two fault systems were recognized: (1) the main NW-trending fault system with mapped fault-length often exceeding 5 km; and (2) a secondary NNE-trending system with shorter faults. The secondary system is parallel to the long axis of the elliptical domal structure of the field. Some of the main faults appear to be composed of en-echelon segments with displacement transfer between the overlapping normal faults (relay faults with relay ramps). The fault systems recognized from the seismic attributes were correlated with well data and core observations. About 13 percent of the fractures seen in cores are non-mineralized. The development of the fault systems was studied by means of clay modeling, computer simulation, and a regional tectonics review. The existing fluid-flow characteristics of individual faults and fractures in the field can be modeled using the present-day stress regime, with the maximum horizontal stress oriented north-northeast. Slip-tendency and dilation-tendency analyses simulating present-day regional stress conditions are indicators of fault and fracture transmissibility. The NNE-striking secondary fault system is parallel to the present-day maximum horizontal stress and could act as a flow conduit in the reservoir.

Present-day stress regimes acting on fault and fracture systems can control where fluid flows within hydrocarbon reservoirs (e.g. Heffer and Dowokpor, 1990). The objective of this study is to accurately map the fault and fracture systems that affect fluid flow in the Thamama reservoir of an offshore United Arab Emirates oil field (Figure 1). Fault and fracture systems occur at many spatial scales, and are grouped here into: (1) seismic scale, (2) subseismic scale, and (3) micro scale.

Seismic-scale faults are routinely imaged with 2-D and 3-D seismic data. Recent improvements, particularly in 3-D seismic acquisition and advanced processing techniques, provide clear images of subtle structural features and facies distribution in hydrocarbon reservoirs. By utilizing a variety of attributes extracted from 3-D seismic data (e.g. dip, similarity, curvature), the subsurface can be further analyzed to estimate the geometry of small-displacement faults (Figure 2).

Subseismic-scale fault and fracture systems, however, are not possible to image using compressional waves alone. These more subtle systems may be indirectly modeled by using outcrop analogs, and physical experiments of both clay and sand materials. These analog models are then calibrated to the actual reservoir by using sensitivity analysis of the experimental settings through numerical simulation. This study used an analog-modeling apparatus consisting of two arms to apply regional stress, and a balloon for dome growth in the central part. These experiments were done at Southwest Research Institute in San Antonio, Texas, USA (D.A. Ferrill, D.W. Sims, A.P. Morris and D.J. Waiting, unpublished JNOC Report 2001).

Micro-scale fractures are usually analyzed using cores and Borehole Image Logs. In our study they were analyzed mainly from core samples (vertical and deviated wells), since the image quality of acquired Borehole Image logs did not allow for their detailed analysis.

The oil producing Thamama Group (Figure 3) in the studied oil field consists of multiple layers of chalky carbonate reservoirs with uniformly poor core permeability, averaging 15 mD, associated with relatively high porosity, frequently above 20 percent. The net-to-gross thickness is estimated to be approximately 300/600 ft. The matrix properties of these reservoirs can be enhanced by a fluid conductive fault/fracture network, and by fracture corridors. The domal structure of the field is likely to be related to a deep-seated salt diapir of infra-Cambrian Hormuz Salt (Figure 4). Previous studies provided evidence that wrench tectonics played an important role in the structural evolution in the region (Marzouk and El Sattar, 1995, Figure 4). At least two distinct tectonic events with different compressive stress directions are interpreted from the features of the Arabian Plate. One is the EW-directed ‘Oman stress’ associated with the Oman Mountain nappes and ophiolite obduction during the Mesozoic. The other is the NNE-directed Cenozoic ‘Zagros stress’. These different stress events reactivated pre-existing basement fault networks, and triggered salt swelling of the infra-Cambrian salt. The salt activation resulted in the structural growth of anticlines above the salt pillows and salt domes in the region. Rapid structural growth at the end of the Middle Cretaceous under the ‘Oman stress’ regime is one of the key factors for the structural doming and faulting of the studied field.

In order to investigate seismic-scale fault geometry, the following seismic attributes were extracted from single 3-D seismic data. (1) Time Dip; (2) Azimuth; (3) Edge Enhancement; (4) Instantaneous Phase; and (5) Curvature (Figure 5, Curvature not illustrated). In this particular oil field, the Time Dip-processed image gives the best visual fault enhancement. The faults were mapped using time slices in each attribute cube every 4 or 8 milliseconds. Lineations observed on time slices were interpreted as seismic-scale faults after being confirmed by other attributes and by vertical seismic cross-sections.

Figure 6 shows an example of fault interpretation on a Time Dip attribute image. Two fault systems are present in this field. The main NW-trending normal faults are commonly traceable for more than 5 km across the structure. They are visible as approximately parallel lineations spaced about 1 km apart in the crest, and as arcuate traces on the flanks of the structure. The secondary NNE-trending normal faults are shorter and less common. The secondary fault system occurs along the long axis of the structure as relatively short fault segments.

NW-Trending Normal Faults (Main Faults)

Figure 7 shows a vertical seismic section with fault traces based on the interpretation of time slices in the attribute cube. Most of the faults are high-angle normal faults belonging to the main NW-trending normal fault system. The maximum fault displacement along the target horizon is just over 140 ft, and the average is less than 30 ft. Horst-and-graben blocks occur in the central part of the structure. Changes in the dip angle of the fault planes are observed in an argillaceous limestone unit that is mechanically more ductile than surrounding beds. This phenomenon can also be seen in outcrops with ductile, typically shale, intervals. Therefore, it might be possible to make lithological predictions from a detailed analysis of the fault geometry. In addition to the normal faults discussed above, a younger fault system that might be related to the Zagros stress is present above the target interval.

A detailed interpretation of the fault geometry was made using the 3-D seismic attributes (Figure 8). The minimum curvature plot along the target horizon provides a detailed image of fault geometries in the field (Figure 8a). Subseismic-scale faults or fracture swarms might be indicated by small curvature anomalies. Close-ups of the main NW-trending normal faults show that what appeared to be one large fault in the overview (Figure 6) is actually a series of smaller en-echelon faults. The overlapping segmented geometry of these faults is indicative of relay faults and the associated relay ramps. Usually relay ramps are bound by a set of faults striking parallel and dipping in the same direction (Figure 8d). Some of the relay ramps are bound by normal faults with opposing dips, creating horst-like structures between the bounding faults (Figure 8c). These are common features in normal-fault systems at all scales, when the faults are arranged en-echelon. In the overlap zone of the en-echelon faults, displacement transfer takes place between the faults. This causes a dip change of strata in the displacement transfer zone that is called a relay ramp (Figure 8c and 8d). Relay ramps are bedding-dip anomalies bounded at each end by cut-off.

Two types of fault propagation can be seen: these are the classical ‘parallel’ type and a curved ‘lateral propagation’ type. A high degree of fracturing that is not necessarily parallel to the main fault trend can be expected in the displacement transfer zone in both types of relay fault geometries.

NNE-trending Normal Faults (Secondary Faults)

The secondary NNE-trending faults are aligned parallel to the long axis of the structure and intersect the main faults approximately at right angles (Figure 6). There is no uniform geometrical intersection relationship between the two fault systems. Some of the secondary faults appear to cut across the main fault trend without relative displacement, whereas other secondary faults appear to abut against the main trend (Figure 6). This secondary fault system corresponds to a similar fault system (at least in the case of its trend) that was seen in the physical analog modeling experiments with clay material (see below).

Clay and sandbox analog modeling experiments were conducted to simulate the interpreted fault systems of the 3-D seismic data. Model simulation included circular and elliptical domes with extension, contractional and strike-slip regional deformation settings. Oblique extension with simultaneous domal uplift was found to yield the best visual analogs with the field data (Figure 9).

Visual comparison between the seismic data (Time Dip image) of the field and the clay experiment shows similarity between both data sets. Fault features such as branching faults, en-echelon normal faults with relay ramps, and laterally curved-fault propagation linked by displacement transfer zones can be seen. Prediction of subseismic-scale fault and fracture geometries is possible based on the analogy of physical model experiments.


The main NW-trending fault system appears to be related to the regional dextral strike-slip deformation, as similar trending faults were observed in nearby fields. The cause of the deformation is the approximate EW-directed ‘Oman stress’ associated with the Oman Mountain nappes and ophiolite obduction during the Late Cretaceous. Salt swelling occurred concurrently with this dextral strike-slip setting. The secondary NNE-trending normal faults and fractures are believed to have propagated along the long axis of the structure, which is perpendicular to the direction of greatest extension during swelling of an elliptical dome. The subsequent exposure of the structure to the ‘Zagros stress’ might have further enhanced this trend or caused propagation of existing fractures and faults.

Fault Statistics

Statistical fault analyses provided insights into the smaller-scale fault geometry of the main fault system. Fault heave and throw were calculated every 250 m along the seismically mapped main faults of the target horizon of the field. Figure 10a is a fault-heave plot along the strike of a typical main fault showing multiple displacement maxima. This indicates the composite nature of the faults, which is possibly related to the joining-up of several, initially en-echelon, smaller faults.

This interpretation is statistically supported by the particularly small fault heave-to-length ratio observed from the seismic data in comparison to outcrop observations (Figure 10b). Typical heave-to-length ratios calculated from seismic data are less than 0.003, which is considerably smaller than the heave-to-length ratios of 0.1 to 0.01 observed in outcrop in West Texas. That indicates that faults in the oil field are very long, but have little displacement when compared with outcrop observations.

To further investigate the composite nature of the faults, the heave-to-length ratio of individual fault segments was analyzed. Fault segments were selected based on the fault-heave plot of individual faults (Figure 10a). Most of the heave-to-length ratios for individual fault segments were again found to be less than 0.01 (Figure 10b), which means that faults in the studied field still have an extreme heave-to-length ratio in comparison with outcrop observations considering the composite nature of faults from seismic data. One likely reason is the limitation of seismic data resolution; that is, the high heave-to-length fault segments might in turn be composed of smaller fault segments that cannot be resolved from the seismic. Another possibility is a strike-slip component that would increase the apparent fault length. Although the seismic data does not show any evidence of strike-slip fault movement in this field, the regional geological setting and core observations could imply some strike-slip movement along the main faults.

Present-day stress influences the flow in matrix reservoirs as well as in fractured reservoirs (Heffer and Dowokpor, 1990; Heffer et al., 1992), and it can also significantly influence the fluid flow along faults and fractures (Barton et al., 1995). Non-mineralized faults and fractures, trending parallel to the maximum principal stress, generally have a higher aperture than those trending perpendicular to the maximum principal stress (Carlsson and Olsson, 1978).

Borehole breakout analyses indicated that the maximum horizontal present-day stress orientation is NNE-SSW in this oil field. It is deduced from core measurements that the maximum horizontal stress also corresponds to the maximum principal stress, with the intermediate principal stress being vertical. Therefore, at present, the oil field is in a strike-slip stress regime.

Geomechanical Interpretation

Slip tendency (Ts) and Dilation tendency (Td) can be used in a semi-quantitative characterization of the fluid-flow behavior of non-mineralized faults and fractures striking in various directions within a given stress field (Morris et al., 1996).


Where: σn = normal stress, τ = shear stress, σ1 = maximum principal compressive stress, and σ3 = minimum principal stress.

Slip tendency is defined as the ratio of shear stress to normal stress on the fault or fracture surface. It depends on the stress field, orientation of the surface and frictional characteristic controlled by rock physical properties. Slip tendency analysis is a way of deciding which faults in an oil field are most likely to have slipped or have current slip potential. Such faults or fault segments could be associated with zones of enhanced fracture permeability. The present-day stress orientation and magnitude (in addition to fluid pressure within the reservoir) control the dilation of faults and fractures, and therefore the tendency for faults to be conductive to flow.

With this method, it was possible to rank the seismically-mapped faults and fault zones in the oil field according to their potential fluid transmissibility. The results indicate strong anisotropic fault transmissibility (Figure 11). Magnitudes of slip and dilation tendency for faults belonging to the secondary fault system are 2 to 10 times higher than those grouped in the main fault system. Assuming that both fault systems are non-mineralized, the results of slip tendency and dilation tendency imply that individual NNE-trending secondary faults have a higher transmissibility than the dominant, primary NW-trending fault system. Further assuming that these NNE-trending faults create discrete fracture corridors with interconnected fracture and fault networks, these results indicated that preferred fluid flow in the direction of the secondary fault systems might exist within this oil reservoir. Circumstantial evidence from interference tests corroborate that such a flow anisotropy exists in this reservoir.

Well Data Indicating Anisotropic Transmissibility

An anisotropic transmissibility of the faults and fractures in this oil field can also be inferred from the fracture interpretation of core samples and from the results of a multi-well interference test. The rose diagram in Figure 12a shows all fracture strikes observed in core samples over the field. NW-trending strike directions are dominant, which is consistent with the main fault trend. There are fewer fractures in the secondary fault direction. Over the sampled interval, 520 fractures were observed of which about 13 percent (65 fractures) were non-mineralized. The rose diagram of non-mineralized (open) fractures observed in core samples is shown in Figure 12b. There are two distinct non-mineralized fracture trends. One is NNE, which parallels the secondary fault strike, and other non-mineralized fractures are parallel to the NW main fault trend.

For the following reasons, a preferential flow through and along the NNE -trending faults and fractures is suggested:

  • 1) The NNE-trending fractures are aligned with the present-day maximum horizontal stress, and therefore have higher dilation and slip tendency than the NW-trending fractures under in-situ stress condition.

  • 2) More than 90 percent of the NW-trending fractures are mineralized, whereas almost all fractures trending NNE are non-mineralized.

  • 3) The NW-trending fractures are thought to be older than the NNE-trending fractures. The NW-trending fractures are most likely created during the Oman Mountain deformation in the Late Cretaceous. The majority of the NNE-trending non-mineralized fractures are believed to be related to, or were reactivated during, the Tertiary Zagros collision. In addition these faults and fractures could have been re-activated during the continuous growth of the domal structure. These faults and fractures will therefore probably crosscut the older NW-trending system. Some of the older NW fractures/faults might also have been reactivated during this time as documented by the rare non-mineralized fracture in this direction.

  • 4) Some of the NW-trending open fractures observed on cores, might actually be closed when under the influence of the maximum principal compressive stress perpendicular to the fracture strike, and may therefore be non-fluid conductive under downhole reservoir stress conditions.

In the target field, multi-well interference tests in wells spaced on average about 2 km were conducted. These showed a strong anisotropic transmissibility with a maximum in a northeasterly direction and a minimum towards the northwest. Assuming an interconnected network of non-mineralized fractures with a fracture network permeability exceeding the matrix permeability, this result is consistent with the theoretical results predicted from the geomechanical modeling.

In this study, two distinct fault systems are described in detail utilizing 3-D seismic attributes. Most of the faults are NW-trending normal faults. They are fairly long and constitute the main fault system, which is associated with predominantly mineralized fractures in cores. This fault/fracture system is interpreted to have originated during the Late Cretaceous Oman Mountain deformation. A secondary fault system, with much shorter faults, trends NNE. It is characterized by mostly non-mineralized fractures. The NNE-trending fault and fracture system parallels the long axis of the elliptical domal structure of the field, and is probably related to the growth of this domal structure and the Tertiary Zagros stress.

The best seismic attributes for fault interpretation are Time Dip, Similarity and Curvatures, which give clear images of small displacement fault geometries. Even though the number of seismic-scale secondary faults is limited, geomechanical modeling indicates that preferred fluid flow is expected through the faults and fractures belonging to this secondary fault system under the present-day NNE-trending maximum horizontal stress and reservoir conditions. The faults are normal faults composed of small fault segments with relay faults and displacement transfer zones. Physical analogs modeling experiments with clay materials and outcrop observations provide us with detailed fault geometry and imply the occurrence of subseismic-scaled faults and fractures.

This paper resulted from work at the Technology Research Center of Japan National Oil Corporation (JNOC/TRC). We thank Abu Dhabi National Oil Company (ADNOC) for permission to publish this paper. We also acknowledge the contributions from the Southwest Research Institute (SWRI) scientists, especially Darrell Sims, David A. Ferrill, Alan P. Morris and Deborah J. Waiting, who carried out physical analog modeling experiments and fieldwork in West Texas. Editing, and the drafting of the final figures were by Gulf PetroLink.


Yoshihiko Tamura is a Senior Geologist with Japan Oil Development Co., Ltd.(JODCO) in Tokyo. He received a BSc in Mineralogy, Petrology and Economic Geology from Tohoku University, Japan in 1985. Prior to becoming a member of the Joint Study Team comprised of Japan National Oil Corporation (JNOC, Currently JOGMEC) and JODCO since 2000, Yoshihiko worked for Sakhalin Oil and Gas Development Company (SODECO) as a Senior Geologist from 1997, and for Japan China Oil Development Company (JCODC) from 1985 to 1993. He is a member of the Japanese Association for Petroleum.


Futoshi Tsuneyama is working on his PhD in Geophysics at Stanford University. Futoshi was a Senior Geophysicist with Japan National Oil Corporation (JNOC) until June 2002. He joined JNOC in 1999 and was a member of the Joint Study Team for the Middle East area. He received a MSc in Geology and Mineralogy from Niigata University, Japan in 1989. Between 1989 and 1998, he worked for Idemitsu Kosan in the upstream division. Futoshi is a member of the EAGE, SEG, SEG Japan and Japanese Association for Petroleum Technology.


Hitoshi Okamura is a Deputy General Manager of Geophysical Department INPEX CORPORATION. He was a Research Project Manager with Japan National Oil Corporation (JNOC, currently JOGMEC). He received a ME in Geology and Mining from Akita University, Japan. Hitoshi joined JNOC in 1984 and has been responsible for the carbonate reservoir characterization project for the Middle East since 1999. Between 1995 and 1998, he worked for Abu Dhabi Marine Operating Company (ADMA-OPCO) as a Review Geophysicist. Hitoshi is a member of the Japanese Association for Petroleum Technology and SEG Japan.


Keiichi Furuya is a Geophysicist with Japan Oil Development Co., Ltd. (JODCO) in Tokyo. He has a BSc (1993) and a MSc (1995) in Earth Science from Chiba University, Japan. Keiichi joined JODCO in 1995 and has a particular interest in reservoir characterization. He is a member of the SEG and SEG Japan.