ABSTRACT

A 3-D geological model of the Kimmeridgian-Tithonian Manifa, Hith, Arab, and Upper Diyab formations in the area of the onshore Central Abu Dhabi Ridge was based on a newly established sequence stratigraphic, sedimentologic, and diagenetic model. It was part of an inter-disciplinary study of the large sour-gas reserves in Abu Dhabi that are mainly hosted by the Arab Formation. The model was used for dynamic evaluations and recommendations for further appraisal and development planning in the studied field.

Fourth-order aggradational and progradational cycles are composed of small-scale fifth-order shallowing-upward cycles, mostly capped by anhydrite within the Arab-ABC. The study area is characterized by a shoreline progradation of the Arab Formation toward the east-northeast marked by high-energy oolitic/bioclastic grainstones of the Upper Arab-D and the Asab Oolite. The Arab-ABC, Hith, and Manifa pinch out toward the northeast. The strongly bioturbated Lower Arab-D is an intrashelf basinal carbonate ramp deposit, largely time-equivalent to the Arab-ABC. The deposition of the Manifa Formation over the Arab Formation was a major back-stepping event of the shallow-water platform before the onset of renewed progradation in the Early Cretaceous.

Well productivity in the Arab-ABC is controlled mainly by thin, permeable dolomitic streaks in the fifth-order cycles at the base of the fourth-order cycles. This has major implications for reservoir management, well completion and stimulation, and development planning. Good reservoir properties have been preserved in the early diagenetic dolomitic streaks. In contrast, the reservoir properties of the Upper Arab-D oolitic/bioclastic grainstones deteriorate with depth due to burial diagenesis.

A rock-type scheme was established because complex diagenetic overprinting prevented the depositional facies from being directly related to petrophysical properties. Special core analysis and the attribution of saturation functions to static and dynamic models were made on a cell-by-cell basis using the scheme and honoring the 3-D depositional facies and property model. The results demonstrated the importance of integrating sedimentological analysis and diagenesis with rock typing and static and dynamic modeling so as to enhance the predictive capabilities of subsurface models.

INTRODUCTION

This paper focuses on an Upper Jurassic intrashelf basin (Figures 1 and 9) and, in particular, on the architecture of the Arab Formation in onshore Abu Dhabi. Previous studies of the Upper Jurassic in Abu Dhabi were mainly of offshore areas or the intrashelf basin infill of the Diyab Formation (de Matos and Hulstrand, 1995; Al-Silwadi et al., 1996; Azer and Peebles, 1998; Al-Suwaidi et al., 2000; Al-Suwaidi and Aziz, 2002). Until now, stratigraphic analysis and well correlations had been mostly limited to the main reservoir-bearing intervals of the Arab depositional system (the Arab-ABC and Upper Arab-D), which has led to confusing definitions of Arab-A, -B, -C and -D. In this study, a large onshore data set was used, including newly acquired information from a well that had been fully cored through the Arab sequence.

The Upper Jurassic Arab-ABC, and Upper Arab-D are hydrocarbon-bearing (Figures 2 and 3) in onshore and offshore Abu Dhabi as well as in neighboring countries (Meyer et al., 2000; Lucia et al., 2001). There is commercial hydrocarbon production from several offshore fields but onshore fields in the Arab Formation have not yet been developed. Reservoirs consist of a series of thinly bedded dolomite and limestone streaks of variable reservoir quality interbedded with impermeable anhydrite layers (Arab-ABC) and a more uniform coarse-grained oolitic/bioclastic limestone unit below (Upper Arab-D). The ultimate top seal for this sequence in western Abu Dhabi was previously thought to be the Hith Anhydrite, but may rather be the tight limestones of the basal Cretaceous Habshan Formation. Evidence for this is that sour gas of similar composition as in the Arab-ABC, has been found in the Manifa Formation (Figure 3) that overlies the Hith Anhydrite in the studied field. Oil and gas with a high H2S content has also been found in the Habshan. This suggests that sour gas has migrated into even higher overlying reservoirs, possibly along fault planes.

A potentially significant sour-oil accumulation has been identified offshore toward the west and north, and adjacent to the large accumulation of sour gas in the Arab Formation within the studied field. This discovery has triggered further exploration activities in the Central Abu Dhabi Ridge area. In the studied onshore field (Figure 1), the Arab-A, -B, -C and -D reservoirs contain gas (Figure 2). The sour gas (57% C1 /C2, 33% H2S, 10% CO2) occurs in a gas-down-to situation. An overall goal of the study team was to develop an optimized sustainable way of separating the hydrocarbon components from the acid gas (H2S, CO2) and use the latter in enhanced oil recovery schemes (not discussed in this paper). A regional-scale 3-D geological field model was developed for the Arab Formation and its surroundings, as well as the results of dynamic modeling from sour-gas production and acid-gas re-injection.

So far, no consensus has been reached on the lithostratigraphic nomenclature of the Upper Jurassic Arab Formation. No model was available that described the sequence architecture of the Formation in Abu Dhabi beyond discussions by Al-Silwadi et al. (1996), Azer and Peebles (1998), Ayoub and En Nadi (2000), and the Arabian Plate summary of Sharland et al. (2001). A prime objective of this study was to generate a comprehensive depositional and high-resolution sequence stratigraphic model. The studied field and its surroundings are ideally suited for this as the area straddles the maximum progradational coastline during Arab times, and therefore allows analysis of lateral as well as vertical facies changes.

In order to allow for quantitative evaluation, the integrated reservoir characterization and modeling study had the following objectives:

  1. To assess key geological uncertainties in the Arab Formation.

  2. To build a regional 3-D static model of the Arab Formation for the quantitative evaluation of reservoir architecture and volumetric ranges.

  3. To provide input for dynamic modeling to allow for uncertainty analysis related to sour-gas production and acid-gas re-injection.

  4. To develop the basic decision-making tools for further Arab appraisal and field development planning.

  5. To provide, in part, for future Arab exploration, appraisal, recovery process evaluation, and field development planning in the Central Abu Dhabi Ridge area.

Using the Arab reservoir as an example, a workflow outline is provided of an integrated reservoir characterization and modeling study as performed by an asset-based team using Shell’s proprietary subsurface 3-D modeling tools, such as GEOCAP and MoReS (Grötsch et al., 2000).

GEOLOGIC SETTING

The Jurassic and Cretaceous sedimentary successions of Abu Dhabi (Figures 2 and 3) fill intrashelf basins that developed as a result of the repeated generation of large volumes of accommodation space on the Arabian Plate (Murris, 1980). Until recently, the organization and architecture of the intrashelf basin infill was poorly understood due to difficulties in dating the events, the limited and low-resolution regional seismic data, and the traditionally used lithostratigraphic terminology (see page 52). Only in the past few years has the complexity and depositional architecture of these individual basins begun to be unraveled with the advent of high-resolution seismic data and the application of sequence stratigraphic concepts. Often the fill of these intrashelf basins contains both reservoir rocks, predominately shallow-water facies along the prograding margins, and source rocks in slightly deeper ramp settings within the central part of the intrashelf basins. However, the quality of both reservoir and source rocks are highly variable through time in the different intrashelf basin infill cycles.

The Arab and its associated formations and members (Hith, Manifa, Arab-ABC, Upper and Lower Arab-D, and Asab Oolite) are of Late Jurassic (Kimmeridgian to Tithonian) age. These lithostratigraphic units are summarized as the Arab sequence. This is a large-scale eastward-prograding intrashelf basinal depositional system that encompasses several sedimentary environments on the Arabian Platform. The Arab-ABC, Hith and Manifa formations pinch out toward the northeasternmost part of the studied field. The Lower Arab-D is considered to be an intrashelf basinal deposit, time-equivalent to the Arab-ABC. Coastal deposits are characterized by oolitic grainstones of the Upper Arab-D and its younger facies equivalent, the Asab Oolite. In onshore and offshore Abu Dhabi, a large-scale eastward progradation of the coastline took place over more than 150 km. This was the expression of the infill of accommodation space in an intrashelf basin generated by a long-term Late Jurassic rise in sea level, which was part of the Tectonostratigraphic Megasequence (TMS) AP7 of Sharland et al. (2001). Slope angles in the intrashelf basin are very low (< 0.5°), which qualifies the depositional system as a carbonate ramp.

Lithostratigraphy

Historically, correlation in the Arab intrashelf basin-infill cycle (Arab-ABC, Manifa, Hith, Arab-D, and Asab Oolite) was primarily based on lithostratigraphic concepts using mainly porosity logs and data from reservoir development for guidance. This caused miscorrelation of reservoir units on both field scale and regional scales (Al-Silwadi et al., 1996). Therefore, all key marker horizons in the Upper Jurassic were re-evaluated with respect to their chronostratigraphic or lithostratigraphic significance for input into the evaluation of the seismic data and the regional 3-D geological model.

The Top Upper Jurassic (Top Manifa) is defined at the base of the high gamma-ray response below the basal Cretaceous limestone (Base Habshan) (Figure 4). It marks the beginning of a renewed transgressive phase and has chronostratigraphic significance. It most likely corresponds to the Sharland et al. (2001) Arabian Plate tectonostratigraphic megasequence AP7/8 boundary at 149 Ma—the Late Jurassic unconformity. The Hith Formation is defined in the model area on gamma-ray markers and not simply on base and top of the anhydrite. This was necessary as the Hith and upper Arab-ABC units become more anhydrite-rich toward the west, and therefore cross time lines.

The Arab Formation is traditionally divided into the Arab-A, -B, -C and Arab-D. However, definitions of Arab-A, -B, and -C vary from field to field and do not necessarily represent the same chronostratigraphic unit. The combined Arab-ABC is here defined as the lithostratigraphic unit composed of high-frequency dolomite-anhydrite shallowing-upward cycles capped by the massive Lower and Upper Hith Anhydrite. The top of the Arab-D reservoir is picked at the top of the clean gamma-ray trend, which is an oolitic and bioclastic grainstone section. It coincides with the occurrence of a black marker limestone that commonly contains pyrite. However, this marker is clearly lithostratigraphic and does not represent a time-line suitable for sequence stratigraphic analysis, although it is important in 3-D model construction (see below). The pick for the Top Diyab Formation is characterized by an increase in the gamma-ray response below the base of the Lower Arab-D (MFS J70 at 152.75 Ma after Sharland et al., 2001). Therefore, the lithostratigraphic unit of the Lower Arab-D is a time-transgressive deposit of a mid- to deeper ramp environment within the Arab intrashelf basin on the Arabian Platform.

Structural History

Pre-Cretaceous structural lineaments are one controlling factor in the development of the present-day large-scale structures in Abu Dhabi. Major lineament sets trend approximately north and northwest.

The N-trending lineaments are basement ridges similar to the Qatar Arch. They cross the whole of Abu Dhabi and are offset by predominantly NW-oriented wrench-fault zones. The studied area is part of one ridge, here named the Central Abu Dhabi Ridge. It was later tectonically overprinted by several anticlines (onshore) and salt domes (offshore) that contain major hydrocarbon accumulations in Abu Dhabi.

The large-scale regional depositional architecture in the Upper Jurassic and Lower Cretaceous of Abu Dhabi shows only small lateral thickness variations. This suggests that the area was tectonically stable while expansion of the NeoTethys was ongoing.

A major phase of tectonic activity in eastern Abu Dhabi occurred in the mid and Late Cretaceous (Aptian to Maastrichtian) in response to the closure of NeoTethys and continental collision with the Arabian Plate. During the deposition of the Middle Cretaceous Wasia Group, the first sign of growth of present-day major structures is indicated by slightly reduced formation thicknesses in several fields, and by thickening in the Hamra syncline of southeastern Abu Dhabi. This was probably the initial response of the foreland to the onset of collision and the downwarping of the eastern Emirates and Oman due to crustal thickening in the Oman Mountains. Progressive infill in this area took place during deposition of the Mishrif, Laffan, and Halul formations (Figure 3) in Turonian to Santonian times. The increased flexing of the foreland area created small extensional faults in a NNW-strike direction toward the end of this period.

By the time the Upper Cretaceous (Campanian) Fiqa Formation (Figure 3) was deposited, the foreland area was being affected by a major compressional phase in response to the emplacement of the Oman Mountains ophiolite thrust sheets. This caused folding and subsequent wrench faulting in Abu Dhabi with the wrench faulting taking place on reactivated, deep-seated WNW-trending lineaments. Wrench faulting and folding in onshore Abu Dhabi and thrusting along the Oman Mountain front ceased at the end of the Cretaceous (Maastrichtian). It had created the anticlinal structures of the present-day major oil and gas fields and had caused crestal erosion and thickening of the Fiqa on their flanks (Figure 5). The complex collision geometry evidenced by the Maastrichtian structures in the Hajar (Oman) Mountains, suggests that several different phases of compression may have taken place. It is possible that the earlier phase was more oblique and created the pull-apart wrench faults prior to compressional wrenching.

3-D seismic data from other parts of the region clearly demonstrate that the large anticlines are cut by wrench-fault zones containing typical small pull-apart and pop-up blocks, and horsetail fault patterns (Javaux et al., 1994; Silva et al., 1996; Melville et al., in press). However, compressional anticlines, as developed in onshore Abu Dhabi, are replaced or overprinted by salt movement and the development of salt domes offshore.

The Maastrichtian Simsima Formation filled part of the residual morphology over the major tectonic structures. It is thicker in the west due to isostatic uplift of the eastern area during the Maastrichtian that caused exposure at the mountain front between Al ‘Ain and Ra’s Al-Khaima. Subsequent burial and uplift, with tilting toward the west, plus compaction and possible ongoing growth of some anticlines due to movement on deep-seated salt cores, continued into the early Tertiary. The Zagros compression in the Oligocene caused preferential tilting down toward the northeast.

Older and deeper structural elements may also have controlled some of the main depositional events. Examples are the edge of the Upper Jurassic Hith Anhydrite and the Asab Oolite with its clinoforms, and the Lower Cretaceous progradation of the Habshan Formation with a north-northwesterly strike direction. The main axis of the Shu’aiba intrashelf basin is oriented northwest and therefore deviates slightly from the general northerly strike direction.

Hydrocarbon System

Over long distances, significant changes in hydrocarbon fill (gas-condensate–light oil) occur in the Arab Formation on the Central Abu Dhabi Ridge. On a smaller scale, the Arab-ABC consists of thinly layered beds of dolomite and anhydrite. Both examples suggest the possibility of lateral and vertical compartmentalization in the reservoir.

The principal source rock is the prolific underlying Diyab Formation, and possibly other unknown sources (Figure 3). The Diyab Formation is directly overlain by the Arab reservoirs (Al-Suwaidi et al., 2000; Al-Suwaidi and Aziz, 2002). It is part of the pre-Arab intrashelf basin infill. However, unlike the Arab Formation, the progradational direction in Abu Dhabi during the Oxfordian to lower Kimmeridgian was from east to west. This is the opposite direction to that in the upper Kimmeridgian to Tithonian Arab intrashelf-basin infill. The Diyab Formation in onshore Abu Dhabi is still within the gas-generating window and the charge may be ongoing.

The Diyab Formation is subdivided into the Upper, Middle, and Lower Diyab lithological units. The Lower Diyab provides more than 90 percent of the Diyab source potential and is the main hydrocarbon-generating interval. It has a very high gamma-ray log response and relatively low sonic velocity. In western Abu Dhabi it is particularly rich in organic matter with a present-day average Total Organic Carbon (TOC) value of up to 1.5 weight percent. However, the original source rock potential of the Lower Diyab must have been substantially higher. The potential of the Middle Diyab is more moderate with an effective source-rock thickness ranging from 25 to 75 ft, with the greatest thickness in the west. The Upper Diyab is organically lean and has a residual TOC of less than 0.8 weight percent in much of Abu Dhabi.

The trapping mechanism for the gas is mainly structural with a stratigraphic component due to the Arab-ABC pinch-out toward the eastern edge of the studied field. Wrench faulting has been interpreted from 3-D seismic data in several onshore and offshore fields in Abu Dhabi and is likely to also occur in the studied field where 3-D seismic acquisition has only recently started. As a third component, it is expected that lateral fault sealing in a ENE-direction supported trapping of hydrocarbons. The Hith Anhydrite and anhydrite interbedded within the Arab-ABC reservoir can form additional intraformational seals or areally limited barriers to vertical flow.

Significant hydrocarbon generation and expulsion started in southwestern onshore Abu Dhabi at the end of the Maastrichtian. The major period of hydrocarbon expulsion was during the early Tertiary in most onshore areas. The present-day maturity of the Diyab Formation indicates that the studied field is still within the gas generation window, as are other onshore structures. However, most of the original source-rock potential is largely exhausted, which is consistent with the relatively low source-rock potential yields recorded from core material. Vertical migration along wrench faults is assumed to be the main transport path for the hydrocarbons, with an additional lateral component being especially important for the Arab reservoirs.

Oil and gas in the Arab are sour due to the high hydrogen sulfide (H2S) content. The H2S was probably formed by thermochemical sulfate reduction (TSR), common in anhydrite (calcium sulfate) successions at temperatures greater than 100°C (Goldhaber and Orr, 1994; Machel et al., 1995; Nöth, 1997; Worden et al., 2000). The current temperature in the reservoir is about 140°C. The products of TSR are H2S, water, and calcite and this basic reaction has two profound implications for the reservoir. Not only are the hydrocarbons sour but, commonly, there is an increase in porosity due to anhydrite dissolution together with occlusion due to calcite precipitation. In general, the H2S content in the Arab Formation in Abu Dhabi increases with the cumulative thickness of anhydrite intercalations, and decreases with shallower depth (less TSR). Therefore, onshore fields in Abu Dhabi tend to have significantly higher H2S contents than offshore areas.

Arab sour gas in the studied field is not fully saturated with elemental sulfur. This may indicate that TSR is still ongoing and mainly constrained by the lack of suitable heavier hydrocarbon components.

DEPOSITIONAL MODEL AND STRATIGRAPHIC ARCHITECTURE

Problems had been recognized in the characterization of the reservoir architecture in the Arab intrashelf basin cycle. A detailed core and sedimentological analysis was therefore made to improve the understanding of the dramatic lateral changes in depositional environments and associated geometries. A high-resolution cyclostratigraphic framework of integrated seismic data, 2-D well correlations, and 3-D geometry modeling was the basis for establishing a predictive reservoir model.

The study of the depositional environments was a crucial step in the integration of microfacies analysis, petrographic core analysis, analog studies, seismic analysis, regional correlation, high-resolution reservoir-scale correlation, and 3-D body modeling.

Facies Analysis

As a result of the facies analysis, it became apparent that the traditionally defined lithostratigraphic units represent depositional systems that are to a large extent laterally time equivalent. The facies belts from west to east (i.e. landward to intrashelf basin) are as follows, based on lithofacies associations (LA):

  • a supratidal to intertidal environment alternating between anhydritic sabkha and salina (LA-1/2; Figure 6a, 6b);

  • an intertidal to subtidal lagoonal environment with algal laminates (LA-3/4; Figure 6c, 6d);

  • a shoreline to inner ramp environment with oolitic and bioclastic grainstones (LA-5; Figure 6e);

  • a mid-ramp environment with a transition from oolitic grainstones to bioturbated wackestones (LA-6; Figure 6e); and

  • an outer ramp environment with micritic, bioturbated limestones (LA-7, Figure 6f).

The lithofacies associations LA-1 to LA-4 form distinct shallowing-upward cycles within fourth-order aggradational and progradational phases in the Arab-ABC (Figure 7a). LA-5 represents the deposits of the prograding coastline during the Kimmeridgian and Tithonian, which cannot be clearly recognized on seismic but are identified on well-log characters (Figure 7b). The younger facies equivalent to this unit is called the ‘Asab Oolite’. LA-6 and LA-7 form the mid- to deeper ramp deposits of the Arab intrashelf basin cycle. Core from Well-X01 suggests that the ramp deposits in the Arab sequence, prograding toward the east, are very similar to those seen in the underlying Upper Diyab sequence that progrades westward. Figure 8 is a schematic block diagram of the lithofacies associations in central Abu Dhabi.

Seismic Stratigraphy and Geometries

As 3-D seismic data are not yet available from the field, the seismic analysis was based on a few 2-D seismic lines of varying vintage and on 3-D seismic from adjacent fields.

The evaluation of impedance contrast and 1-D synthetic seismograms from several wells suggests that the top Arab-D or its lateral equivalent, the top Asab Oolite, is the main mappable reflection in the Arab succession in central Abu Dhabi. The acoustic response can be explained by a negative impedance contrast between the tight lime-mudstones of the lowermost Arab-ABC and the porous Upper Arab-D oolitic grainstones (15–50 m thick). Laterally, the transition is from the tight basal Cretaceous limestones of the Habshan Formation to the porous Asab Oolite. However, the seismic interpretation is complicated, as the amplitudes created at these boundaries are rather weak due to the low average velocity contrast between the Arab-ABC and the Upper Arab-D. Nevertheless, results show that the main seismic reflection at the Arab level is a lithostratigraphic boundary not a chronostratigraphic boundary and this hampers a sequence stratigraphic analysis based only on seismic.

The acoustic impedance inversion on a NE-trending 2-D siesmic line across the field (Figure 10) highlights the presence of the porous Asab Oolite zone in the northeast. It also suggests that low-angle clinoforms occur in the Lower Arab-D.

2-D synthetic seismic was generated using the same seismic section, together with the depositional geometries to be expected from the sequence stratigraphic model (see below) and the first iteration of the 3-D static model. The interpretation of the original 2-D seismic line and the 2-D synthetic seismic modeling using porosity data from wells and the model, supported the presence of low-angle eastward-prograding clinoforms in the Lower Arab-D, indicating the infill of the intrashelf basin morphology. Sensitivity tests with varying degrees of noise suggested that prograding clinoforms could be imaged, despite low-porosity contrasts in the Lower Arab-D.

The 2-D synthetic seismic was also used to investigate the imaging of small-scale, NS-oriented ridges of oolitic grainstones in the Upper Arab-D and the younger facies-equivalent of the Asab Oolite. The synthetic seismic model was constrained based on observed thickness variations in the Upper Arab-D from core (15–50 m), and velocity contrasts from the well data. The ridges, recognizable from 3-D seismic surveys in Abu Dhabi (Figure 9), are visible on the 2-D synthetic seismic (Figure 10) and give support to the depositional model and its sequence stratigraphic interpretation. The ridges are oriented north-northwest, parallel to the paleocoastline of the Arab intrashelf basin (Figures 1 and 9). Thus, although the seismic data is sparse, it supports the interpretation of the reservoir architecture.

A comparison of the 3-D seismic amplitude extraction maps with the flattened 2-D seismic data along the intrashelf basin margin indicated a good match with the ridge geometries and the spacing observed between the ridges (Figure 10). This suggested that the ridge spacing in Central Abu Dhabi was recording high-frequency cyclicity and trends in progradational and aggradational phases of the accommodation space infill within the intrashelf basin. In addition, the data on ridge spacing can be used as a quantitative input for predictive 3-D modeling (Figure 11), as petrographic analysis (discussed below) indicated that the reservoir quality in the Upper Arab-D was a function of the abundance of ooids.

SEQUENCE STRATIGRAPHY

The Arab intrashelf basin-infill cycle (including Upper Arab-D, Lower Arab-D, Arab-ABC, Manifa, Hith, and Asab Oolite) is a major depositional system that progrades from west to east through central Abu Dhabi (Figure 12). This is in the opposite direction to the intrashelf basin cycle of the underlying Diyab Formation (de Matos and Hulstrand, 1995; Al-Suwaidi et al., 2000). In a chronostratigraphic framework, the dolomite-anhydrite shallowing-upward cycles of the Arab-ABC are time-equivalent to the Upper Arab-D oolitic grainstone belt. This formed the progradational coastline during the Kimmeridgian and Tithonian and its trend can be derived from the seismic data (Figure 9). The Lower Arab-D (equivalent to the Fahahil Formation) forms the time-equivalent ramp deposit leading into the intrashelf basin center.

Following the maximum flooding surface (MFS) based concept of Sharland et al. (2001), approximately 330 m of Arab were deposited between MFS J70 (152.75 Ma) and MFS J110 (147 Ma) within about 5 to 6 my (see Figure 4). The two flooding surfaces can be considered as reliable correlation markers in the United Arab Emirates and surrounding countries.

The MFS J80, J90, and J100 of Sharland et al. (2001) could potentially coincide with three aggradational cycle sets in the Arab-ABC. Their initial fifth-order shallowing-upward cycles within the fourth-order cycles contain dolomite streaks that have the best reservoir properties in the Arab-ABC. However, due to the lack of biostratigraphic and isotope stratigraphic resolution, such regional comparisons of candidate flooding surfaces requires further validation in order to avoid miscorrelation between fields as a result of lateral facies changes.

It is difficult to correlate with confidence the time-equivalent fourth-order cycle packages of the shallow-water Arab-ABC with the intrashelf basin interior ramp setting of the Lower Arab-D. This is because of the intervening coastal deposits of the Upper Arab-D and the limited well data available from the Lower Arab-D. However, as the 3-D reservoir model is based on a combination of sequence stratigraphic and lithostratigraphic units best representing major flow units, this was not a requirement (see below).

Cyclostratigraphy

The predominantly intertidal to supratidal environments of the Arab-ABC show a distinct high-frequency cycle-stacking pattern (see also Azer and Peebles, 1998). It is a reflection of alternating aggradational and progradational periods, and it is therefore a function of changes in the accommodation space through time related to the intrashelf basin infill (Figure 12).

The shallowing-upward cycles have a predominantly dolomitic intertidal facies at the base and are capped by playa or sabkha-type anhydrite, often with signs of subaerial exposure. Diagenetic overprinting of individual cycles can affect the underlying sediment, and cause cycle amalgamation.

Correlation of fourth-order cycle stacking patterns based on Fisher plots (Fischer, 1964; Sadler et al., 1993) using the approach of Day (1997), indicated that the best reservoir quality (highest permeability) was preserved at the base of the fourth-order cycles immediately above the main thick anhydrite intercalations. This showed that the increase in reservoir quality in the Arab-ABC apparently coincided with the maximum increase in accommodation space of the fourth-order cycles during aggradation (Figure 13).

The upper two good-quality reservoir intervals in Figure 13 may be related to MFS J100 and J90 of Sharland et al. (2001: Fig. 4.43) respectively. The lower reservoir is between MFS J90 and J80, and is support for the observation that best reservoir quality in the study area is related to re-flooding of the platform and to an overall aggradational fourth-order phase.

From seismic data it was also apparent that the distance between coastal oolitic ridges in the Upper Arab-D varied greatly. Closely spaced ridges indicated higher rates of aggradation; conversely, widely spaced ridges indicated periods of faster progradation. Hence, both the cycle stacking pattern and the seismic data reflected relative changes in sea level as recorded in the Arab-ABC and Upper Arab-D, and their resultant reservoir quality.

Lithostratigraphy versus Sequence Stratigraphy

The above discussion has illustrated that lithostratigraphic units are to a large degree time-equivalent lateral depositional facies within the Arab intrashelf-basin infill cycle. In general, the thick anhydritic units in the Arab-ABC and the Hith represented periods of limited accommodation space and of cycle amalgamation, and therefore of rapid progradation. Hence, the Hith Anhydrite represented a period of maximum progradation (low accommodation space) with its lateral equivalent, the Asab Oolite, forming the paleocoast environment. The last stage of the Arab intrashelf basin cycle was represented by the Manifa Formation. This unit is interpreted as a drowning succession of the Arab cycle that culminated in a major backstepping of the platform and the onset of the first intrashelf-basin infill cycle of the Cretaceous Habshan Formation. Hence, the position of the sequence boundary between AP7 and AP8 of Sharland et al. (2001: Fig. 4.43) is supported by the observations in central Abu Dhabi.

DIAGENESIS

Petrographic analyses were made to characterize the nature and timing of diagenetic overprinting of the reservoir and to allow prediction of the distribution of pore-enhancing and pore-destructive phases. At the same time, the relationship between depositional facies and diagenetic overprinting was investigated. The diagenetic analysis also provided the framework for the development of the reservoir rock-type scheme and the input for the 3-D static and dynamic model. It is apparent from these studies that diagenesis has exerted a strong secondary control on reservoir quality within the Arab, Hith, and Manifa formations.

Paragenesis

The inferred paragenetic sequence for the Arab in the study area is shown in Figure 14 with the key events summarized below:

Anhydrite Formation and Dolomitization

This occurred during, or shortly after, deposition in an evaporitic setting and led to the formation of laterally rather continuous anhydrite and interbedded dolomite layers in the Arab-ABC (Butler, 1969). The best reservoir quality is in dolomite streaks that occur in the basal part of the fourth-order cycles (Figure 15a).

Pore-filling Calcite Cementation

Interparticle pores in the oolitic and bioclastic grainstones of the Upper Arab-D are commonly blocked by porosity-destroying non-ferroan calcite cements (Figure 15b). The intraparticle pore-filling cementation phase was associated with an initial charge of hydrocarbons, as suggested by residual bitumen within the blocky calcite, and is therefore of burial origin.

Leaching of Allochems

Leaching in the Upper Arab-D postdates the cementation by burial-type calcite and is a function of ooid abundance in the beach deposits. The leaching event predated a last-stage precipitation of pore-filling corrosive cement that reduced the leached porosity in the ooid coatings. This final cementation occurred late in burial history and is assumed to be a result of thermochemical sulfate reduction on anhydrite to generate calcite and water (Machel et al., 1995).

Principal Diagenetic Controls on Reservoir Quality

The primary depositional fabric of the sediment, and ultimately of the reservoir architecture, can exert a strong control on reservoir quality, especially within the Lower Arab-D clinoforms. However, throughout most of the Arab reservoirs, diagenetic modification has significantly affected reservoir quality. In order to classify the changes, a petrophysically based rock-type scheme was required alongside the lithofacies association scheme that could only classify depositional facies (Figure 16). The following diagenetic processes are considered the most important for control on reservoir quality.

Dolomitization

LA-3 (and to a lesser extent LA-4 and LA-7) is pervasively dolomitized. Depending on the dolomite fabric, this can lead to either an increase or reduction in reservoir quality, as follows:

  • The higher permeability dolomites, mainly in LA-3, generally have crystal-supported planar-e fabrics (Mazzullo, 1992) with moderately well-connected intercrystalline macropores (>10%; see Figure 15a). However, where intercrystalline pores are occluded by residual limestone and/or anhydrite and non-ferroan calcite cements, permeability is reduced substantially (Figure 6c).

  • Planar-s fabrics, with high proportions of interlocking dolomite crystals, have a lower permeability due to the blocking of intercrystalline microporosity and/or microfractures.

It is assumed that accommodation space was restricted and water depths during deposition were shallow enough to permit such pervasive dolomitization of the Arab-ABC host sediment. As accommodation space was further reduced later in the third-order highstand systems tract (HST), replacement of the sediment by nodular anhydrite became increasingly common, so destroying porosity and degrading reservoir quality.

Reflux of brines with elevated salinities and/or flood recharge and evaporitic pumping in a sabkha setting were potential mechanisms for dolomitization of the Arab-ABC during the third-order mid-HST (Adams and Rhodes, 1960; Hardie, 1987; McKenzie, 1991). Due to the deep-water setting of the dolomite units in the Lower Arab-D, it is probable that there was a second mechanism for dolomitization as a result of the diffusion or circulation of seawater at depositional surfaces.

Calcite Cementation and Subsequent Leaching

The interparticle pore system of the Upper Arab-D grainstone lithofacies association (LA-5) has been destroyed by the precipitation of blocky or drusy non-ferroan calcite cements (Figure 15b). Permeability (0.01–2 mD) is therefore controlled by the degree of microporosity hosted by peloids and ooids, and also by intercrystalline boundaries.

In summary, dolomitization is the dominant diagenetic controlling factor for porosity preservation within the Arab-ABC cycles, whereas calcite cementation and subsequent leaching has determined reservoir quality within Upper Arab-D. However, depositional facies cannot be directly related to reservoir quality due to incoherent diagenetic overprinting within the reservoir (Figure 16). From the available core material in the studied field, no obvious depth-dependant trend in reservoir quality was recognized.

RESERVOIR DEVELOPMENT

The main factors that controlled reservoir development in the Arab Formation were three distinctly different elements within the lithostratigraphic units Arab-ABC (equivalent to the Qatar Formation) and Arab-D (equivalent to the Fahahil/Jubaila).

  1. Thin streaks of early diagenetic dolomite within the small-scale, shallowing-upward cycles of the Arab-ABC pinch out toward the east-northeast in the studied field.

  2. Oolitic and bioclastic grainstones of variable thickness occur in the Upper Arab-D. In them, most of the porosity is represented by intraparticle micro-porosity within oolitic coatings. Hence, the reservoir quality is largely controlled by the abundance of ooids.

  3. Good-quality stratiform dolomite streaks in the Lower Arab-D (mid- to outer-ramp setting) enclosed in tight bioturbated limestones (mudstone to wackestone) of the gently dipping carbonate ramp.

Reservoir Quality

The two main controls on reservoir quality are (1) cyclostratigraphy and (2) a complex series of diagenetic events.

  1. The best reservoir quality in the lagoonal and sabkha deposits of the Arab-ABC was developed during the early phases of the generation of accommodation space in fourth-order cycles (Figure 13, 15a). This occurred when the shallow-water platforms were flooded (basal fifth-order cycle in fourth-order stack). Reservoir properties deteriorated as the accommodation space was reduced.

  2. The diagenetic events that affected Arab reservoir quality most significantly are, in approximate paragenetic order:

    • replacive dolomitization—variably porosity reducing and enhancing (LA-3, LA-4, rarely LA-7);

    • anhydrite and gypsum formation—porosity reducing (predominantly LA-1/2);

    • pore-filling and allochem-replacive non-ferroan calcite cementation—porosity reducing (predominantly LA-5);

    • intraparticle leaching of allochems—porosity enhancing (predominantly LA-5); and

    • syn-depositional/shallow-burial dolomite characteristic of LA-3 and LA-4 (Hith and Arab-ABC) and rarely bed tops in LA-7 (Lower Arab-D).

Reservoir Rock-type Scheme

As reservoir quality and properties cannot be directly related to depositional facies/architecture, a petrophysically based rock-type scheme was developed. This was for use with open-hole logs and for input into static and dynamic modeling whereby saturation functions were attributed on a cell-by-cell basis.

As discussed above, individual lithofacies associations have highly variable reservoir properties. As such, they cannot be used to predict permeability distribution or to attribute saturation and relative permeability functions to a 3-D reservoir model (Figure 16). In order to reconcile this issue, a petrophysically based rock-typing approach was followed so as to improve the understanding of reservoir-quality distribution in the Arab sequence.

The scheme is that of Grötsch et al. (1998). A preliminary rock-fabric scheme, based on petrographic and core observations was established and used to select representative samples for Mercury Injection Capillary Pressure (MICP) analysis. The preliminary scheme was refined using MICP curves and associated pore-throat size distributions to develop the petrophysically based rock-type scheme. After several iterations, five reservoir rock types—one anhydrite and four carbonates—were defined, each with distinctive pore characteristics that could be related to the sedimentary fabrics and diagenetic overprinting (Figure 17).

The five Reservoir Rock Types (RRT) have the following characteristics:

  • RRT1 (Figure 17a): good-quality planar-e dolomite with moderately well-connected intercrystalline macroporosity—porosity = 21%, range 17.5–29.2%; permeability = 10 mD, range 2.87–49.5 mD.

  • RRT2 (Figure 17b): moderate-quality microporous lime grainstones (cemented interparticle pores) and planar-e/s dolomite—porosity = 13%, range 8.1–18.1%; permeability = 0.7 mD, range 0.20–3.48 mD.

  • RRT3 (Figure 17c): poor-quality microporous lime wackestones, packstones and grainstones, and planar-s dolomites—porosity = 7%, range 5.0–12.0%; permeability = 0.1 mD, range 0.02–0.30 mD.

  • RRT4 (Figure 17d): negligible-quality lime mudstones and wackestones and rare planar-s dolomites and tight peloidal skeletal grainstones, non-reservoir—porosity = 1%, range 0.01–14.2%; permeability = 0.01 mD, range 0.01–0.05 mD.

  • RRT5 (Figure 17e): negligible-quality anhydrite, non-reservoir—porosity = 1%, range 0.01–7.4%; permeability = 0.02 mD, range 0.01–0.28 mD.

This scheme was then upscaled (Figure 18) so that the rock types could be predicted from open-hole logs and particularly from neutron (NPHI) and density (RHOB) logs. This was achieved by extrapolating the rock-type samples from the combined petrographic and MICP dataset of 60 sample from three wells, to the total petrographic dataset of 205 samples from four wells and subsequently to the total routine core-analysis dataset. The core analysis rock-type dataset and lithofacies association distribution was used to assign rock-types to the open-hole log data for the cored intervals.

To allow saturation functions to be attributed to the 3-D reservoir model, high case, most likely, and low case saturation curve scenarios were defined for each rock type. These were later used to constrain Gas Initially In Place (GIIP) uncertainty.

STATIC RESERVOIR MODEL

The main goals of the static (and dynamic) reservoir modeling were to build a 3-D geological model for the Arab sequence in order to allow evaluation of uncertainties relating to reservoir architecture, GIIP, sour-gas production, acid-gas re-injection, and field-development planning. Pre-studies and computer modeling demonstrated the complexity of the Arab reservoirs and their sequence stratigraphic architecture (Figure 12).

Hierarchical Correlation

In order to allow for realistic model construction (including object-based modeling), a detailed and hierarchical digital well-correlation scheme was a pre-requisite. All correlations were performed using GEOLOGIX, which is a module in the 3-D modeling suite GEOCAP.

Main Correlation Units

The definition of the main markers and reservoir units (first hierarchical level) used in the 3-D model was based on a combination of sequence stratigraphic and lithostratigraphic units. The constraints that were given to realistically model the reservoir architecture as shown in Figure 4 used only 10 wells in the large model area of 70 by 50 km. Figure 4 shows a pronounced streakiness in the Arab-ABC reservoir.

High-resolution Correlation

As a second step in reservoir modeling, a high-resolution well correlation (second hierarchical level) was performed, based on the sequence stratigraphic architecture of the main correlation units—the Manifa, Hith, Arab-ABC, Upper Arab-D, and Lower Arab-D (Figure 19). The correlation allowed 3-D modeling of all the main reservoir elements within each correlation hierarchy (e.g. pinch out of Manifa, Hith, and Arab-ABC; thin streaks of dolomitic reservoir in Arab-ABC in Figures 20 and 21; and oolitic grainstones of the Upper Arab-D).

Both first and second hierarchical levels were used for sequentially modeling the reservoirs. As a result, the second-level models were constrained by the object shapes of the first level.

Structural Model

The studied field has not been surveyed by 3-D seismic. However, the interpretation of the 2-D seismic results suggests that wrench faults are present in the modeled area. Recent horizontal appraisal and development drilling in several locations also confirmed the faults and associated fracture systems.

This indicated that the field is transected by a series of wrench-fault zones, similar to other onshore fields to the north, south, and east, which have been surveyed by 3-D seismic (Figure 9). Three main structural directions have been identified based mainly on Formation Micro-Imaging (FMI). The two most prominent directions are N135°E and N105°E (Figure 5).

However, the location of the faults cannot be determined with a high degree of certainty, and fault mapping is not possible based on the available well information and the sparse 2-D seismic data. Therefore, faults have been delineated using data from several sources, such as the regional stress field evolution, 3-D seismic from neighboring fields, pressure maps, water production, step-like porosity changes in reservoirs, horizontal well trajectories, and FMI to prepare a schematic vertical fault network.

The presence of even small offsets along faults becomes important in terms of dynamic-flow behavior. This is particularly so in the Arab-ABC reservoir where the juxtaposition of its small-scale dolomite-anhydrite cycles has caused localized sealing or communication between different layers. A tentative and schematic faulted top-structure map was therefore generated for input into the 3-D model so as to allow a sensitivity analysis in the static and dynamic models.

As there were few 2-D seismic lines available in the field, the top marker horizon of the model was interpolated from Cretaceous Thamama markers derived from the many well penetrations. Subsequently, isopachs were drawn down to the Top Jurassic level using the 10 deep-well penetrations in the area of interest. The 3-D geological model was constructed in such a way that it can be quickly updated with a new 3-D seismic-derived structural map, once it is available.

Palinspastic Reconstruction

The top structure map (see Figure 5) was used in GEOCAP as the reference for a palinspastic reconstruction. In this case, the marker horizon used in the model was MFS J110 of Sharland et al. (2001) near to the base of the Lower Cretaceous Habshan Formation, which was seen in all wells. In part of the model it equates with the Top Manifa Formation but to the east it is the Top Asab Oolite. The marker can be identified by its high gamma-ray response in all wells of the Central Abu Dhabi Ridge (Figure 4) and regionally (Sharland et al., 2001: Fig. 4.43). This horizon was chosen as the Arab-ABC and Manifa pinch out in the eastern part of the model area.

Palinspastic reconstruction of the structured grid and flattening on the Top Jurassic allowed successive object modeling of depositional bodies in a paleodomain based on the hierarchical correlation.

Object Modeling

Based on well data and high-resolution hierarchical well correlations, depositional geometries were modeled in 3-D following the same hierarchy as used for correlation (Figure 19). Most were not present throughout the whole of the area modeled but are correlatable, as they are present in more than one well. Examples are the Arab-ABC (first hierarchical level), and the dolomitic reservoir streaks (second hierarchical level). The sum of the depositional objects in the individual levels determined the reservoir architecture and therefore the 3-D depositional facies model (Figures 2022a).

Depositional Facies Model

Object modeling based on facies analysis of core and log data allowed definition of the Arab depositional system in the Central Abu Dhabi Ridge area. It formed the basis of the 3-D reservoir architecture that combined sedimentological data, sequence stratigraphic analysis, and the definition of the main flow units (Figure 22a). The high-resolution object-based depositional facies model and the structural models were the key constraints in reservoir property modeling.

Property Modeling

3-D property distribution for porosity and permeability was to a large degree constrained by the 3-D depositional facies model. Kriging, directional kriging along depositional trends, and co-kriging using trend maps as constraints were the preferred interpolation methods. Kriging orientations were preferentially parallel to the prograding shoreline of the Arab intrashelf basin infill and trended north.

Because the model area was large and the available well control was limited, multiple realizations were generated for various properties to cater for the remaining uncertainties. All realizations were based on a deterministic control derived from the depositional facies analysis, regional geologic studies, and diagenetic analysis. Such external control was used for guiding stochastic modeling as well as for pure deterministic modeling.

The vertical resolution for property modeling was chosen as 0.3 m (1 ft) in order to honor the thin-streak nature of the Arab-ABC reservoir and to allow calculation of the vertical permeability by upscaling. The latter was a crucial input parameter for dynamic modeling.

  • Porosity. Based on the depositional facies analysis and the 3-D body modeling results, a base-case porosity realization was prepared. This used parallel coastline trends for the Arab-ABC and Upper Arab-D and a down-flank property-reducing trend for the limestones of the Upper Arab-D together with the petrophysical control from 10 wells (Figure 22b). A low-case realization was generated by co-kriging with a porosity map using downscaling factors and by maintaining data integrity at the well locations. Upgrading the net pore volume by 20 percent generated the high-case realization. Ranges used for upgrading and downgrading 3-D property distributions were based on uncertainty estimates derived from the petrophysical re-evaluation of the Arab log data.

  • Permeability. Permeability data were available from many core plugs as 90 percent of the wells in the Arab Formation have core data available for the modeling interval. It was therefore not necessary to apply the inherently uncertain permeability estimation based on log data. The core data provided a good basis for 3-D permeability modeling that was well constrained by the 3-D reservoir architecture and the coastline orientation (Figure 22c).

Reservoir Rock-type Model

Petrographic analysis demonstrated that there was no clear relationship between depositional facies, reservoir properties, and saturation functions. A new petrophysically based rock-type scheme was therefore developed and used as input for the 3-D static and dynamic model.

The reservoir rock-type model presented an additional modeling step between depositional facies modeling and property modeling (porosity and permeability) on the one hand and saturation modeling and dynamic simulation on the other (Grötsch et al., 1998). Whereas the main flow units are defined by the structured hierarchical facies model, which are layer or body based, the rock-type model had to be cell based (Figure 22d). This was essential in order to allow attribution of different saturation-height functions during saturation modeling and model initialization, on a cell-by-cell basis within simulation layers.

Rock types were attributed to each cell by using the porosity and permeability models and definitions from the petrographic study. This study had identified the five rock types graphically represented in Figure 18. On a porosity-permeability cross-plot, four distinct areas were defined using petrographic data and capillary pressure measurements. Attributing these relationships to the cellular property models allowed for gradual lateral changes in saturation-height/relative permeability functions within the particular reservoir architecture and within each individual facies body (e.g. down-flank rock-type deterioration in Upper Arab-D). The cell-by-cell based rock-type model was then used as input for saturation modeling and the initialization of the dynamic model (Figure 22d).

Saturation

Results from a first-pass model and the petrophysical review showed large uncertainties when using log saturation data for 3-D interpolation. Uncertainties were the result of calculated log saturations as well as spatial property distribution using various interpolation techniques. A key constraining factor was the number of wells available and their location with respect to the 3-D reservoir architecture. Therefore, saturation-height functions together with the developed rock-type model were used for saturation modeling in the static model and for model initialization in the dynamic model (Figure 23a).

Saturations were calculated per rock type per cell, as a function of height above Free Water Level (FWL) using the newly acquired mercury capillary pressure data and the resulting saturation-height functions. Connate water saturations were defined for each rock type and honored in the saturation functions. To cater for the uncertainty ranges in the data set, three realizations (Low, Mid, High) with one type curve each were defined (Figure 23b).

All Arab wells have a gas-down-to situation, and therefore the FWL for the main reservoir units had to be estimated. The estimate was based on the dominant rock type in the reservoir unit, saturation-height functions for the dominant rock type, uncertainty range in the saturation-height functions, water saturation ranges measured from logs, and matching saturations from the 3-D model, together with log-derived saturations at well locations. This provided a minimum FWL estimate that was the basis for further FWL definitions for each realization of the saturation model (Figure 24). Synthetic saturation logs from the model suggested that the deeper FWLs better matched with actual resistivity log data and were therefore more realistic.

Upscaling

Preservation of key geological features in the dynamic model proved essential for prediction of the dynamic behavior within the Arab reservoir. This was achieved by upscaling in the vertical sense only, while preserving the tentative fault scheme. In addition, the geological model based on sedimentary objects/bodies facilitated the preservation of reservoir heterogeneity that was crucial for fluid flow during upscaling and elsewhere in the dynamic model. All dolomitic reservoir streaks of the Arab-ABC were preserved as individual layers (the reservoir parts of fifth-order cycles), separated by non-reservoir anhydrite layers. This allowed preservation of small-scale shallowing-upward cycles and the modeling of body pinch-out. The Upper Arab-D is the only unit that is continuous over the total model area. For upscaling, it was subdivided into four reservoir simulation layers that reflected the observed (rather than a uniform) property distribution. Arithmetic averaging was used for vertical porosity upscaling, and harmonic averaging was applied to the upscaling of the permeability.

DYNAMIC MODEL

The static model is an integral part of field development planning. It formed the basis for the analysis of well-test data, the design of appraisal-well testing, and the analytical evaluation of productivity and injectivity. A dynamic element model was extracted to evaluate the feasibility of acid-gas re-injection, and of a full-field model to determine well- and in-field production compression requirements under various development scenarios.

Well-test Analysis

Test data from appraisal wells were inconclusive due to the limited duration of the tests and because of mechanical failures during testing. Well-test data from the most recent down-flank appraisal well (Well-X01; Figure 5) were analyzed in detail. Additional information from production logging showed that only a single Arab-ABC dolomite streak controlled well inflow, which proved crucial in the realistic analysis of well-test data. Furthermore, production logging indicated the importance of effective well-stimulation design and acid diversion.

The analysis was performed both analytically and by dynamic element modeling. Through this iterative process, a flow match of the test results was achieved by using linear flow with parallel faults. The results suggested that the reservoir was faulted in the vicinity of the well and that faults impaired lateral fluid flow in the Arab-ABC. In combination with analog data and inflow performance relationships derived from the crestal Well-X16 and midflank Well-X27, production tests were used to assess the potential benefit of horizontal wells in the Arab reservoirs. It was concluded that the absolute open-flow potentials of 110 to 120 million standard cubic feet/day (MMscf/d) could be achieved from horizontal well trajectories. For appraisal and development, horizontal drilling is planned such that the horizontal sections will penetrate the full vertical section of the Arab reservoirs and link all Arab-ABC shallowing-upward cycles to the wellbore.

Well-test Design

An appraisal well is planned to confirm the sustainable output from such a high-angle well. Other objectives of well testing are to ascertain reservoir boundaries and to measure reservoir parameters such as permeability, skin, and reservoir pressure.

The layered nature of the Arab-ABC added to the complexity of the well-test design and analysis. Anhydrite layers are likely to act as vertical flow barriers. Therefore, even though the thickness of the Arab-ABC averaged 52 m, it is possible that the effective reservoir is composed of only a few layers with very limited cross-flow. Assuming the well will penetrate the entire vertical section of the Arab-ABC reservoir, each separate layer is likely to be vertically isolated and the effective well length in each layer would be much less than the total length. In such complex cases, analytical models and solutions would not satisfactorily capture the physics of reservoir behavior. Hence, a single well 3-D dynamic model in MoReS was used to capture the complete spatial and physical reservoir description. The reservoir pressure was then exported to the PanSystem™ for well-test analysis.

Two major questions had to be addressed: (1) How long should the test last? and (2) What is the optimum test strategy? (i.e. whether to test the Arab-ABC alone or commingled with the Upper Arab-D). Results showed that testing the Arab-ABC and Upper Arab-D together would not give proper information on faulting or compartmentalization of the reservoir. Testing the Arab-ABC reservoir streaks alone reveals faults more easily and this result allowed the test objectives to be refined. Determining the distances to possible faults is the reason for extended testing of the horizontal section in the Arab-ABC, whereas the prime objective of testing the Upper Arab-D alone, and commingled with Arab-ABC, was to establish the overall well deliverability.

Feasibility of Acid-gas Re-injection

The development of the Arab sour gas in the studied field requires the flexibility to re-inject acid gas (80% H2S and 20% CO2) into the Arab reservoirs. Given the layered nature and reservoir property contrasts of the Arab-ABC, the issue of long-term reservoir management had to be addressed. It is expected that future acid-gas injection will take place at the crest of the structure because of the favorable reservoir properties found in appraisal wells, the requirement of limiting acid-gas contamination to only part of the field, and health and safety considerations such as the proximity to surface facilities. The purpose of this feasibility study was to assess the displacement process and the recovery factor of methane in various acid-gas disposal scenarios. For this purpose, an element model was extracted from the static model to investigate the feasibility of simultaneous production of sour-gas and re-injection of acid-gas.

A symmetrical element model of dimensions 1,500 by 1,500 m (Figure 25) was generated that reflected vertical reservoir property contrasts as observed in the mid-dip appraisal Well-X27. The model consisted of one injector/producer pair positioned diagonally across the model. Given the layered nature and reservoir property contrasts of the Arab-ABC—as opposed to the relatively homogenous Upper Arab-D reservoir—it was determined that acid-gas injection (if necessary) should be limited to the Upper Arab-D, and that injection into the Arab-ABC should be delayed as long as possible.

It was concluded that no special reservoir management policy was required to maximize the recovery of sour gas should re-injection of acid gas become necessary. Although it was recommended that re-injection should be avoided, it may be necessary for a limited amount of time because of operational circumstances. Completion of the wells should allow selective monitoring and optional shut-in of the Arab-ABC and/or Arab-D reservoir units. Monitoring surrounding producer wells is a requirement for the detection of potential premature acid-gas breakthrough.

Full-field Modeling

In addition to element models, a full-field model was extracted from the geological model so as to support field-development planning (Figure 26). The dynamic model was limited to the main field and was slightly refined compared to the static model in order to allow for the modeling of horizontal wells. Faults were transferred from the static model and they separate the dynamic model into 11 blocks (Figure 5). Flow across the faults is assumed if the juxtaposition of reservoir versus non-reservoir layers permits.

In the Arab-ABC, each of the carbonate/dolomite reservoir streaks was represented by a separate simulation layer or body of grid blocks that were separated by the anhydrite layers/bodies. The latter were subsequently voided out for flow modeling. Permeabilities and rock types had been previously defined from the static model. The rock type selected for each simulation grid cell is that which occupied the largest net pore volume in the static model. This rock type was then used to assign the capillary pressure and relative permeability to each grid cell in the model as defined in the rock-type scheme. The model was initialized, assuming proven contact levels for gas-initially-in-place.

Several development scenarios were investigated to determine the requirements and timing of development wells and in-field production compression. A phased development is planned with a maximum off-take rate of 710 MMscf/d. Based on the currently available data for the static and dynamic modeling results and the Arab reservoir property trends, the development requires a higher than expected number of wells. Several alternatives were investigated, which indicated that a flexible approach to reservoir development was crucial, and that the appraisal value of early wells positioned both on the crest and flank of the structure was significant. Appraisal results may lead to a larger project with a maximum off-take rate of 910 MMscf/d. Several crestal producer wells are planned and will be completed early in order that they can be converted to acid gas injectors at minimum cost and short notice.

SUMMARY AND CONCLUSIONS

The following regional trends were observed in the Arab sequence:

  • Arab-ABC increases significantly in thickness from east to west and pinches out at the eastern edge of the investigated field.

  • Reservoir quality in the onshore areas is poorer than in the Arab offshore, as related to increased diagenetic overprinting caused by the increased depth of burial and thermochemical sulfate reduction.

  • Well data suggest a complex variability of lateral reservoir quality within the Arab-ABC shallowing-upward cycles. No simple down-flank decrease in properties is observed in the Arab-ABC in the study area whereas this occurs in the overlying Cretaceous Thamama reservoirs (Grötsch et al., 1998; Melville et al., in press). Large lateral variations occur within individual cycles or stacks of cycles.

  • Early diagenetic (syn- to early post-depositional) dolomitization of Arab-ABC reservoir streaks enhances the likelihood of porosity preservation during burial.

  • The reservoir quality of the Upper Arab-D oolitic and bioclastic grainstone is likely to deteriorate in a down-flank direction due to burial diagenesis.

The new geological model for the onshore part of the Central Abu Dhabi Ridge was prepared for the Manifa, Hith, Arab, and Upper Diyab formations. It was built with a high vertical resolution in order to tackle the crucial issue of well inflow being predominantly controlled by thin streaks of enhanced reservoir properties in the Arab-ABC. The prime advantage of this model is its preservation of the reservoir architecture with its strong vertical and lateral permeability contrast.

The modeling exercise has demonstrated the complexity of the Arab reservoir and its sequence/cyclostratigraphic architecture. In the studied field, clear signs of a westward progradation of the Arab Formation coastal deposits (oolitic/bioclastic grainstone in Upper Arab-D, Asab Oolite) have been identified. The Arab-ABC, Hith, and Manifa formations pinch out in the northeastern part of the field, whereas the Lower Arab-D is an intrashelf basinal deposit that is to a large degree the time-equivalent of the Arab-ABC.

The results of the integrated reservoir characterization and modeling exercise show that well productivity in the Arab-ABC is predominantly controlled by the development of thin dolomitic streaks in a series of hierarchically organized small-scale, shallowing-upward cycles. This is important for well completion, well stimulation, and development planning, as the best reservoir properties are preferentially found at the base of fourth-order aggradational cycles.

The models presented have been greatly improved by using iterative feedback and updating as a vital part of the modeling process. The full-field geological model allows for upscaling and input from various dynamic modeling processes (element, sector, full-field) and the evaluation/optimization of development scenarios.

The full integration of the various detailed models and a multidisciplinary approach are of paramount importance for reservoir characterization, modeling, and development planning for the complex gas development of the Arab Formation. Understanding the remaining uncertainties and constraints has led to a clear definition of requirements with respect to project phasing, appraisal activities, well completion, reservoir management, and monitoring.

ACKNOWLEDGMENTS

The authors thank the Abu Dhabi National Oil Company and Shell Abu Dhabi BV for permission to publish these results. In particular, we would like to express our gratitude for the support of Mohamed Al-Qubaisi, Mohammed Juma, and Ali Al Shamsi. We also thank Cathy Hollis of Badley Ashton and Associates (now with Shell) for her support of this work. The comments of GeoArabia’s anonymous referees were greatly appreciated. The design and drafting of the final figures was by Gulf PetroLink.

ABOUT THE AUTHORS

Jürgen Grötsch is Reservoir Studies Team Leader for the Al Furat Petroleum Company in Damascus, Syria. He joined Al Furat in late 2002. He was previously Geology and Geophysics Co-ordinator at Shell Abu Dhabi assigned to the Abu Dhabi National Oil Company. Jürgen obtained his MS from the University of Erlangen, Germany in 1987. After being a Research Fellow at Scripps Institution of Oceanography, San Diego, he received a PhD from the University of Erlangen-Nürnberg in 1991 for a study on the evolution of Cretaceous carbonate platforms. Following a post-doctoral fellowship at the University of Tübingen, Jürgen joined Shell International Exploration and Production as a Seismic Interpreter. Later, he was assigned to Shell Exploration and Production Technology and Research as a Production Geologist. There, he worked mainly on the application and development of novel 3-D modeling techniques for carbonate reservoirs with emphasis on the integration of 3-D seismic and outcrop analogs. Jürgen has provided technical services to operating companies in Oman, Abu Dhabi, Venezuela, Kazakhstan, Malaysia, and the Philippines. He has given numerous keynote speeches and lectures in international conferences around the world and is an editor and reviewer for several international journals.

j.grotsch@afpc.net.sy

Omar Suwaina is Strategy and Exploration Team Leader with the Abu Dhabi National Oil Company (ADNOC). He was awarded a BSc in Geological Engineering from Colorado School of Mines in 1990 and joined ADNOC the same year. Omar has led and worked with several multidisciplinary exploration teams, and has been involved in the processing and interpretation of 2-D and 3-D seismic surveys in Abu Dhabi. His main interests are in seismic reservoir characterization and exploration. He is a member of the Emirates Society of Geoscience and SEG.

osuwaina@adnoc.com

Ghiath Ajlani is Team Leader of the Geophysical Technology Support Group in the Abu Dhabi National Oil Company (ADNOC). He has a BSc in Geology from the University of Damascus, a BSc in Geophysics from the University of Missouri-Rolla, and an MSc in Geophysics from the University of Utah, Salt Lake City. Before joining ADNOC in 1996, Ghaith was a Geophysical Advisor for Conoco from 1984 in North America and the Middle East. His major areas of interest are seismic modeling, inversion, land and marine multidisciplinary geophysical feasibility studies, seismic acquisition and processing, and reservoir geophysics for oil and gas exploration and development. Ghiath is a member of the SEG.

gajlani@adnoc.com

Ahmed Taher is a Senior Geologist with the Strategy and Exploration Team of the Abu Dhabi National Oil Company (ADNOC). He has a BSc from United Arab Emirates University. He joined ADNOC in 1982 as a Well-site Geologist and became an Explorationist in the Areas Section in 1986. During 1997 and 1998, Ahmed worked as a member of the Zakum Development Team. Ahmed has professional interests in stratigraphic trap and basin modeling evaluation and has published several technical papers. He is a member of the Emirates Society of Geoscience.

ataher@adnoc.com

Reyad El-Khassawneh is a Senior Geologist with the Abu Dhabi National Oil Company (ADNOC). He received a BSc degree in Geology from Mousel University, Iraq, in 1976 and a PhD in Geology from Bucharest University, Romania, in 1980. Reyad has been involved in numerous exploration and production projects in Abu Dhabi, and until recently he was working with the ADNOC/Shell Sour Gas Development Team.

Stephen Lokier is a Carbonate Reservoir Geologist with Badley Ashton and Associates. He obtained a BSc in Geological Sciences from Oxford Brookes University in 1996. He then undertook research at Royal Holloway College, University of London on the influences of clastic sediments in a variety of equatorial carbonate depositional settings. Since gaining his PhD in 2000, Stephen has been employed by Badley Ashton on several Middle Eastern reservoirs and has coordinated the development of the company’s Carbonate Sedimentology course. Stephen is a fellow of the Geological Society of London and a member of AAPG, BSRG, IAS, PESGB, and SEPM. He has presented papers at many international conferences. Stephen is particularly interested in the application and integration of sedimentological and petrological techniques so as to enhance paleoenvironmental analysis, and sequence stratigraphic and diagenetic interpretations, thereby aiding the construction of reservoir architecture.

slokier@badley-ashton.co.uk

lokier@blinternet.com

Gordon Coy is a Carbonate Reservoir Geologist with Badley Ashton and Associates. He obtained a BSc in Geology from the University of Birmingham in 1994 before moving to the Department of Earth Sciences, University of Cambridge, to undertake research on regional dolomitization of the Akhdar Group of Oman. After completing his PhD in 1998, Gordon joined Badley Ashton where he has worked on several Middle Eastern reservoirs. He has been responsible for the development of Microsoft Access-based petrographic databases, and currently acts as Middle East (Abu Dhabi) Team Leader. Gordon is a member of the BSRG, PESGB and SEPM. He has presented papers at national and international conferences and has published on diagenesis and dolomitization. Gordon has interests in carbonate depositional systems, diagenesis, pore-system analysis and rock typing, and their integration into reservoir models.

gcoy@badley-ashton.co.uk

H. (Erik) van der Weerd is a Reservoir Studies Coordinator with the joint Abu Dhabi National Oil Company/Shell Sour Gas Team. He received a Petroleum Engineering and Geophysics degree from the Technical University of Delft, The Netherlands, in 1979 and joined Shell International the same year. Erik has been assigned to various Shell operating units including Spain, Oman, Denmark, Venezuela, The Netherlands, and now the United Arab Emirates. He is a member of the SPE.

evanderweerd@adnoc.com

erik.vanderweerd@shelldub.simis.com

Shehadeh Masalmeh is a Reservoir Engineer with Shell. He holds a BSc (1990) in Physics from Birzeit University, Palestine. He obtained an MSc (Honors) in Atomic Physics from the University of Amsterdam in 1992 and a PhD in Laser Physics from the University of Leiden in 1997. Shehadeh joined Shell the same year to work in the Special Core Analysis Team. He is a member of the Society of Core Analysts and has published several articles covering SCAL, wettability, hysteresis, and transition zones. Shehadeh’s research interests include multiphase flow in porous media, wettability, hysteresis, enhanced oil recovery processes, and asphaltene precipitation and its effect on sweep efficiency.

smasalmeh@adnoc.com

shehadeh.masalmeh@shelldub.simis.com

Johan van Dorp is Team Leader of a joint Abu Dhabi National Oil Company/Shell Sour Gas Study Team. He has an MSc (1980) in Experimental Physics from RU Utrecht and joined Shell in 1981. After a spell with Shell Expro in North Sea operations, Johan moved to NAM as a Reservoir Engineer. This was followed by assignments in California and Denmark. Johan transferred to Shell Abu Dhabi in 1999. He is a member of the SPE and is professionally interested in recovery mechanisms, well testing, PVT, and process analysis.

jdorp@adnoc.com

johan.j.j.van-dorp@shelldub.simis.com