ABSTRACT

The Safah oil field was discovered in 1983 on the north-plunging Lekhwair Arch of northwest Oman. The arch lacks any significant structural closure and the accumulation is stratigraphically trapped within chalky high porosity-low permeability Upper Shu’aiba carbonates of mid-late Aptian age. The complexity of its trapping geometry, internal reservoir architecture, reservoir quality and hydrocarbon charge history precluded easy explanation and geological models used to describe the field evolved quite significantly over time to accommodate new data and changing regional perspectives. These had a profound influence, first upon the decision to test what was a speculative new concept exploration prospect and later during appraisal and development, in defining an optimum static reservoir model, history matching and efficient field management strategies.

The original play concept developed out of a loosely constrained regional structural and stratigraphic synthesis. Early isopach mapping had identified an enormous paleohigh on the North Lekhwair Arch, which appeared well placed to receive charge in the later Cretaceous and early Tertiary. This was tilted northward during the late Miocene, when any structurally trapped oil or gas must have been spilled to the south. However, nearby analogs suggested that the northeastern margin of the Upper Shu’aiba intraplatform Bab Basin crossed the arch in the vicinity of the paleohigh and it seemed possible that remigrating hydrocarbons might have been stratigraphically trapped against the impermeable basinal facies equivalents of Shu’aiba platform carbonates. Safah-1x was drilled to test this hypothesis, just to the north of the weakly defined Upper Shu’aiba shelf break. It encountered a thin pay zone at the northern end of what proved to be a more than 1 billion barrels STIIOP accumulation.

The complexity of the field became increasingly apparent during appraisal drilling. Both differentiated shelf-to-basin and layered mid-shelf ramp depositional models were proposed to describe its unexpectedly heterogeneous internal reservoir architecture. Independent petrographic, fluid property and oil isotope analyses seemingly contradicted more likely stratigraphic correlations and consensus on a static reservoir model proved difficult to reach. As a result, geologically simple layered reservoir descriptions were favored during the early development of the field. However, as the regional perspective improved with better local analogs and increasing amounts of well and seismic data, attention eventually refocused back toward a more sophisticated stratigraphic explanation. The reservoir is now interpreted to be a low-energy mid-Shu’aiba highstand composite sequence with younger lowstand shales and offlapping carbonate shoals to the south. The updip trapping mechanism is far more complex than originally anticipated, formed by discontinuities between the porous lowstand shoals. The enigmatic relationship between stratigraphic architecture and in-reservoir PVT fluid properties and d13C isotope gradients appear to reflect dual charging by a high GOR Jurassic-sourced oil during the late Cretaceous-early Tertiary and low GOR Silurian oils in the Miocene. Internal stratigraphic baffles prevented complete homogenization and the PVT and isotope gradients remain as geochemical palimpsests.

This resolution of initially rather contradictory observations was achieved by synthesizing data into a coherent narrative logic, most consistent with the available geological information at all scales, from the regional and general to the local and specific. Although more advanced seismic, petrographic and geochemical technologies certainly encouraged increasingly precise interpretations, the issues they raised were still geological and so still most effectively utilized within the context of such narratives. Ultimately, it was only by assessing these against broader geological perspectives that it proved possible to judge the validity of in-field interpretations with any confidence.

INTRODUCTION

We contend that successful geological predictions made from incomplete information depend upon recognizing repeatable geographic, spatial or chronostratigraphic trends and patterns. Predictive success at any scale from regional to local or specific, can only come from testing, reinforcing and modifying working interpretations against a broader global or regional perspective. We will attempt to demonstrate this thesis by tracing the evolution of geological ideas and concepts developed over the life of the Safah field in Oman, from a weakly constrained new concept play in 1980 to the mature field it is today. Throughout this period, a wide variety of often-contradictory ideas and interpretations were proposed. Many were initially rejected because of competing, more fashionable hypotheses, only to be rediscovered and applied later in a different form. Although multiple working hypotheses or probabilistic distributions can describe such geological uncertainty, it does not resolve ambiguity. Based upon our experience at Safah, we will argue that ambiguities can only be removed by extrapolating from wider, more regional geological perspectives and analogies rather than relying upon geographically limited data sets. Not only does this provide a template with which to judge the relative importance of seemingly contradictory information, but it also serves to constrain predictions and define the limits between reasonable projections and outright speculation.

The Safah field is located in Block 9 in north-central Oman, close to the border with Abu Dhabi (Figure 1). The current license is the residual part of the originally much larger Suneinah Concession awarded to Quintana in 1976. Gulf joined the venture as operator in 1977 and Occidental in 1979 (Figure 2). Exploration first focused on the Safah area in 1980 to 1981, following several marginal to unsuccessful exploration tests farther to the east. The Safah-1x well was drilled to evaluate a speculative new-concept stratigraphic play on the north-plunging North Lekhwair Arch, downdip from the regionally prominent Lekhwair High and the crestal Lekhwair field (>200 million barrels—MMbbl) (Figures 1 and 3). The initial well was unpromising as it produced only 300 barrels of oil a day (bopd) from a high-porosity and low-permeability chalky carbonate reservoir in the upper part of the mid-Aptian Shu’aiba Formation.

Occidental took over the operation of the field in early 1984 and subsequently drilled 13 wells to appraise the find. These established the full areal extent of the field, confirming the stratigraphic nature of the trap and the very marginal quality of the reservoir (average 250–600 bopd/well). Uncertainty about its commercial viability, compounded by the 1986 oil-price collapse, led to a 14-month drilling moratorium. Eventually, with Oman Government encouragement, it was decided to go forward with development. Gulf’s 35 percent equity interest in the field was absorbed by Chevron in 1986 and sold to Neste in 1991. Occidental continued as operator (65% interest) and drilled more than 120 vertical development wells between 1987 and 1991 which flowed at rates of between 40 to 1,000 bopd/well. Later infill wells were predominantly horizontal with significantly better flow rates ranging between 1,000 and 3,000 bopd/well. Over this period, field production increased steadily from 2,000 bopd in 1985 to 25,000 bopd in 1991. Early attempts to boost recovery started in 1991 with a pilot gas-injection program. This was expanded significantly in 1993 and increased production to over 30,000 bopd. The first 3-D seismic survey was shot over the southeastern part of the field in 1994. Results were initially disappointing but improved considerably with time and by 1998 good quality 3-D had been acquired over the entire southern part of the residual concession. A pilot waterflood program was begun in 1999 with initially very promising results. Over the last few years, field production has varied between 32,000 and 35,000 bopd with 155 million standard cubic feet of gas/day (scfd) and 8 to 10,000 barrels (bbl) of water/day.

The field is geologically complex and several aspects of its stratigraphy, structure, and hydrocarbon charge history have proved difficult to resolve. These include:

  • Trap and Reservoir Architecture. The oil is trapped by overlying Albian Nahr Umr shales and a southwesterly facing Upper Shu’aiba shelf-to basin-facies transition that straddles the north-plunging North Lekhwair Arch (Figure 3). It was originally anticipated that a distinct shelf-margin break would form the critical updip lateral seal, but subsequent appraisal drilling demonstrated that the accumulation extended farther to the south, trapped by poorly defined discontinuities within the basin sequence of shoal-water limestones and shales. This is still favored as the primary trapping mechanism, although the degree of connectivity between the thin-bedded limestones remains unclear. Subtle low-displacement cross-arch faults have also been suggested as a contributory factor, but their continuity and sealing effectiveness is uncertain.

    The reservoir interval is comparatively thin and of low relief with complex internal geometry, composed of several high-frequency depositional sequences. The absence of distinctive internal marker horizons and poor bio-stratigraphic resolution, combined with apparently conflicting pressure-volume-temperature (PVT) and oil-isotope gradients have compounded this complexity, and made it difficult to achieve a lasting concensus about the field’s internal architecture. Even now, complete agreement remains elusive.

  • Reservoir Quality. The reservoir is dominated by fine-grained moldic and chalky micro-intercrystalline dissolution porosities averaging 22 percent, but with low permeabilities of less than 5 mD. The origin of the microcrystalline fabric and its subsequent dissolution have been very contentious issues and are still unresolved with arguments for sub-aerial meteoric, shallow and deep burial diagenesis. Consequently, it has proved difficult to develop a sufficiently robust geological model capable of accurately predicting permeability distribution within the reservoir.

  • Hydrocarbon Charge. Similarly, the origin of the oil in the reservoir remains in dispute. While this is now generally recognized to be a mixture of two geochemically parents, there is still disagreement about their origin. Several possible sources have been proposed including organic-rich facies in the Infracambrian (Huqf and Q oils), Silurian (Sahmah/B), Jurassic (Hanifa and Diyab) and basinal mid-Shu’aiba (Figure 3).

The charge history also awaits definitive resolution. The accumulation lies at the southern end of a very large paleostructural closure, now in a downdip position on the north-plunging North Lekhwair Arch. This developed as part of a peripheral bulge during the emplacement of the Semail-Hawasina allochthon in the late Santonian to early Campanian. It was well placed to receive hydrocarbon charge throughout the later Cretaceous and early Tertiary. Closure was finally dissipated by northerly tilting during an Early to mid-Miocene plate collision when the Oman Mountains were uplifted and partially exhumed (Figures 1 and 3). As presently understood, the paleoclosure was first charged before the Miocene tilting and then again during (and perhaps shortly after) the tilting episode, when significant stratigraphic closure developed against the Upper Shu’aiba shelf- to basin-facies transition immediately to the south. Confirmation must wait for additional drilling on the crestal part of the paleohigh farther north.

These several contentious issues have had a very profound influence upon static reservoir models, field development strategies and hydrocarbon reserve assessments (Figure 4). Many attempts have been made to resolve them and a variety of interpretations have been proposed, considered, defended and sometimes rejected over time, becoming more sophisticated with increasing well and seismic control, better local analogs and a wider, more profound regional perspective. Although there are still some areas of uncertainty remaining, we believe we have a sufficiently detailed appreciation of the field’s complexity with which to measure the accuracy of our earlier geological predictions about trap, reservoir and charge. We will follow these conceptual threads from the inception of the original play concept to their current resolution, emphasizing those deductive approaches that seem to us to have been most useful in unraveling the field’s geological history. Of course, much will be unique to Safah and we recognize that modern analytic techniques might make some of our earlier tentative steps redundant. Nevertheless, while play, prospect and field descriptions are increasingly constrained by ever more sophisticated technology, we hope to demonstrate that broad conceptual perspectives continue to offer the most effective way of assessing geological predictions on every scale from regional to local.

EXPLORATION (1980 to 1983)

At the time that Occidental joined the Suneinah joint venture, the company was attempting to identify a few attractive big-reserve exploration projects from a large pool of international opportunities. A relatively small exploration portfolio allowed for consistent project selection and qualitative risk assessment by a centralized, geologically literate senior management. Exploration effort focused upon opportunities in areas with working petroleum systems already established or strongly implied from nearby information. The Suneinah Concession in Oman (Figure 1) fitted these criteria and compared well with other competing opportunities. The license area was extremely large with only a few wells, just to the north and east of several giant and super-giant accumulations in a seemingly excellent hydrocarbon environment. Previous operators had already acquired a large amount of seismic data and two plays were immediately apparent:

  • Several small low-relief structural closures in the foreland, similar to those tested earlier by the Shams-1x, Shams-2, and Malik wells in the license itself and by several wells in the adjoining area (Figure 1).

  • Far more speculative and seismically unconstrained possibilities along the flanks of the Oman Mountains.

The foreland play was too small to be of more than marginal interest and the company focused its attention on the mountain front trend where there seemed to be an acceptable chance of finding analogs of the giant Fahud and Natih oil fields just to the south. This potential was considered enough to justify joining the venture. By that time, it was already apparent that the Lekhwair High and its north-plunging structural nose lay in an attractive regional position for hydrocarbon charge. However, in the absence of any significant structural closure, it remained only a tantalizing possibility and exploration efforts were focused on the central and eastern part of the concession. Wadi Rafash-1x (WR1X, Figure 1) was drilled in 1981 on a low-relief foreland structure similar to that tested earlier by Shams-1x, and the Suneinah-2x (1982) well was targeted at a mountain front prospect, with only very limited success. Enthusiasm for the project began to wane, but the discovery of a hydrocarbon-bearing, rudist-dominated interval in Wadi Rafash-1x (Frost et al., 1983) was sufficiently encouraging to justify a regional stratigraphic review of the Shu’aiba Formation. This ultimately re-focused exploration attention back toward the North Lekhwair area.

Trap and Reservoir Architecture

Several possible trapping mechanisms were considered in an attempt to define a drillable prospect on this seemingly featureless structural nose. These included:

Cross-arch Faulting Play Concept

The quality of seismic data was strongly influenced by longitudinal sand dunes with over 500 ft relief, separated by flat gravel plains. Severe seismic processing problems associated with the dunes were mirrored by linear, poor data zones at the top Shu’aiba reflector. These hinted at the possibility of low-displacement cross-arch faults capable of forming three-way dip fault traps. Some support for this hypothesis came from their possible continuity with an arcuate array of well-defined fault strands splaying out northwestward across the Concession from the Natih-Fahud area. Regional structural analysis suggested that they had developed as a result of stress dissipation at the termination of the Maradi transcurrent fault zone (Figure 1), and might have formed viable traps during the later Cretaceous, immediately prior to oil migration and charge. However, the concept was generally considered contrived because of the coincidence of the interpreted faults with the surface dunes and did not find much support.

Diagenetic Trapping Play Concept

Structural mapping of the western part of the Concession in 1981 and 1982 had revealed the presence of a very large paleoclosure on the North Lekhwair Arch (Figures 1, 3 and 5). Regional structural analysis of the Oman Mountain Foreland region indicated that the closure had developed as part of a regional peripheral bulge during the late Santonian to early Campanian. This occurred in response to the obduction and emplacement of an allochthonous oceanic wedge from the northeast (Glennie et al., 1974; Robertson and Searle, 1989; Boote et al., 1990). It appeared to have remained in closure until the Early to mid-Miocene, when it was dissipated by northerly tilting during a phase of regional structural readjustment associated with transpressional uplift of the Oman Mountains.

Following earlier speculation by Dunnington (1967), this observation suggested the possibility of a diagenetic trap. Dunnington (1967), Harris et al. (1968) and Twombley and Scott (1975) had all noticed significant down-flank loss of porosity, permeability and stratal thickness in the Bab and Bu Hasa fields of onshore Abu Dhabi, commensurate with increasing pressure solution, cementation, and stylolitization. The structural closure of both fields had developed in the Late Cretaceous to early Tertiary (Hajash, 1967; Twombley and Scott, 1975) and the reduced crestal lithification was attributed to diagenetic inhibition by early hydrocarbon charge. Near-original reservoir porosities of 25 percent preserved within the hydrocarbon-saturated intervals in the two within structures, contrasted with porosities of 12 percent or less in the water legs below. The two fields also experienced a late Tertiary phase of structural tilting associated with the Zagros orogeny, in some ways analogous to that on the North Lekhwair Arch. However, this appeared to have been far milder with little or no observable effect on their oil-water contacts. Johnson and Budd (1975) also recorded a similar pattern of downflank stylolitization and lithification in the nearby Asab field.

The phenomenon provided a useful analogy for the Safah diagenetic trapping concept. As at Bu Hasa and Bab, the North Lekhwair paleoclosure appeared well positioned for charge throughout the later Cretaceous and early Tertiary. It was argued that subsequent post-charge diagenesis might have reduced the permeability below the paleoclosure sufficiently to prevent spillage and dissipation during later Miocene tilting. However, this found little support as it was considered unlikely that the shallow Shu’aiba target would have experienced severe enough diagenesis to have created a geologically effective lateral and bottom seal.

Upper Shu’aiba Shelf Margin Facies Pinch-out Play Concept

Regional well control available at the time had demonstrated that the upper Shu’aiba sequence in the central and northern part of the Concession was dominated by shelf carbonates with impermeable basinal facies to the southwest. The transition zone between the northern shelf and southern basin offered the chance of a viable stratigraphic trapping mechanism somewhere in the larger North Lekhwair Arch area (Figure 6).

A useful regional perspective supporting this concept was provided by several excellent reviews of the Shu’aiba Formation elsewhere in the region. The earliest of these were by Hajash (1967) and Harris et al. (1968) who described a possible mirror image Upper Shu’aiba shelf-margin sequence in Abu Dhabi at the Bu Hasa field which passed northward into dense basinal Shu’aiba facies of the Bab field (Figure 6). Harris et al. (1968) recognized three distinct phases of shelf-margin development.

  • Phase 1: a regionally continuous Lithocodium-Bacinella boundstone and chalky mudstone ramp sequence (Shu’aiba A Member).

  • Phase 2: initial differentiation between deep-water basinal mudstones to the north and a low-relief algal platform in the south, dominated by Lithocodium-Bacinella, with a narrow transitional shelf-derived wackestone to packstone apron.

  • Phase 3: development of a higher relief shelf margin with a reef core dominated by caprotinid rudists, miliolid and rudist wackestones and packstones to the south, with forereef rudist grainstone and packstone detritus, and more basinal time-equivalent Orbitolina wackestones, carbonate mudstones and shales to the north. Initially referred to as the ‘Dense Shu’aiba Limestone’ or ‘Shu’aiba Equivalent’, this more basinal facies was later designated the Bab Member by Hassan et al. (1975).

Although the stratal architecture of the shelf-to-basin transition was not described in any detail, the accompanying illustrations suggested that the algal platform sequence and early part of the buildup phase was dominated by vertical aggradation, evolving into a more progradational margin toward the end of the Shu’aiba with several offlapping shoals (Harris et al. 1968).

By 1975, enough exploratory wells had been drilled so as to extend this north-facing shelf-to-basin transition in a near-linear trend across central onshore Abu Dhabi (Twombley and Scott, 1975). Murris (1980) published the first comprehensive regional Upper Shu’aiba facies map of the southern Arabian Platform in 1980. He defined the Lower Shu’aiba A Member as a laterally extensive ramp carbonate passing up into a mid-late Aptian differentiated carbonate shelf developed in response to a relative rise of sea level. His map revealed the full extent of the Upper Shu’aiba intraplatform basin centered on Abu Dhabi, but with a pronounced re-entrant reaching southeastward into Oman (Figure 6). Farther outboard within the intraplatform basin, locally developed shallow-water shoaling sequences were also encountered at Zerkouh and Mandous in offshore Abu Dhabi and thinner coralgal facies at onshore Jarn Yaphour and Zibara (Hassan et al. 1975), all apparently founded on salt-cored bathymetric highs (Figure 6).

This work provided a critical template for extending and enhancing the very limited regional well control available within and adjacent to the Suneinah Concession. Upper Shu’aiba platform, shelf margin and basinal facies were tentatively identified from gamma-ray well-log profiles calibrated against the Bu Hasa analog and proprietary lithofacies information. Hassan et al. (1975) had previously suggested that the limits of porous Upper Shu’aiba shelfal carbonates could be defined empirically by thicknesses of 350 ft or more. Using this as a guide, the widely spaced well information was combined with Murris’s mid-late Aptian facies map to project a south-facing Upper Shu’aiba shelf margin obliquely across the North Lekhwair Arch (Figure 6). Its position was constrained still further by subtle Shu’aiba thickening observed on a few key seismic lines in the extreme southern part of the paleohigh (Figure 5). Viewed in its regional context, there seemed a reasonable possibility that this might represent the predicted shelf-break geometry and transition into impermeable basin facies, well placed to provide a stratigraphic barrier across the Arch and capable of damming hydrocarbons re-migrating updip toward the south during Miocene tilting.

Reservoir Quality

Attempts at predicting the quality of the reservoir along the postulated shelf margin proved difficult because of the wide range of Upper Shu’aiba porosities and permeabilities in nearby wells. Wadi Rafash-1x with its thick caprinid-rich rudist shoal sequence offered the nearest local analogy (Frost et al., 1983). However, porosity was rather poor (11% average), irregularly distributed, and largely occluded by mud and early diagenetic micritic cements. This contrasted strikingly with more porous shelf carbonates in Malik-1x. As neither well appeared to be representative of a shelf margin type sequence, the Bu Hasa reef trend was ultimately favored as the more realistic analog (Figure 6).

Detailed descriptions of the Bu Hasa reservoir sequence by Hassan et al. (1975) and Twombley and Scott (1975) once again provided an useful analog. They reported excellent reservoir properties with both primary and secondary porosities and permeabilities of up to 30 percent and 500 mD or more. Although the field as a whole appeared to be volumetrically dominated by chalky microcrystalline mud matrix (Hassan et al., 1975), the more grain-rich facies had excellent primary intergranular and framework porosities.

Postdepositional leaching was recognized as the most significant porosity forming process, particularly in sediments rich in unstable aragonitic skeletal debris (Hassan et al., 1975; Twombley and Scott, 1975). This secondary porosity appeared non-facies selective and affected the Shu’aiba reservoir uniformly throughout the field. It was considered to have developed shortly after deposition (Twombley and Scott 1975) and was attributed to a period of regional emergence at the end of Shu’aiba deposition (Harris et al., 1968; Hassan et al., 1975; and Twombley and Scott, 1975). However, direct evidence of exposure was ambiguous with only some minor local truncation (Harris et al., 1968) and reworked phosphatic limestone pebbles collected in shallow depressions on the bored top Shu’aiba surface (Twombley and Scott, 1975). Farther north in the Bab and Asab fields, even these features were absent and a dense basinal Bab facies appeared to pass directly up into Lower Albian Nahr Umr shales without significant break.

Upper Aptian ammonites had been found in the Bab Member of the Bab field (Harris et al., 1968) and the implied stratigraphic continuity with the Aptian-aged shelf carbonates suggested that exposure must have been of relatively short duration and confined to the platform. Nevertheless, while perhaps of only minor importance in Abu Dhabi, Powers et al. (1966) and later Murris (1980), Entsminger (1981) and Soliman and Shamlan (1982) all recognized a very profound top Shu’aiba unconformity eroding down to Devonian and older rocks in central Arabia and southern Oman. Consequently, it was argued that the severe uplift and erosional unroofing implied by this event was sufficient to have encouraged regional meteoric fluid flow basinward toward the east and might have been responsible for the dissolution and porosity enhancement observed at Bu Hasa.

Extrapolations from this regional perspective suggested that similar primary and secondary porosities might reasonably be expected on the North Lekhwair Arch within the analogous high-energy, rudist-dominated reservoir facies anticipated there. Significant pressure solution, stylolitization, and lithification were not expected because of the shallow depth of the target.

Hydrocarbon Charge

There was early consensus that the North Lekhwair Arch was optimally positioned for migration from the northeast, north and northwest, updip toward the Lekhwair and Mender fields in the south. Several possible source rocks had been highlighted by earlier work. Clarke (1975) favored Thamama Group intraformational shales as the origin of oils in the Thamama reservoir in central and eastern Abu Dhabi. Murris (1980, 1981) later distinguished two primary source intervals: the upper Oxfordian/lower Kimmeridgian intraplatform basinal facies of the Hanifa or Diyab Formation and the Bab Member of the Shu’aiba. He indicated that good quality Hanifa source facies (2–6% total organic carbon) was restricted to Qatar and the western part of Abu Dhabi where it had charged both Jurassic and Lower Cretaceous reservoirs (Murris, 1980; Schlumberger, 1981). However, its quality appeared to deteriorate toward the south and east and he suggested that the Bab intraplatform sequence was the more probable source for Thamama accumulations in eastern Abu Dhabi. Consequently, it was argued that this was the most likely candidate to charge the North Lekhwair Arch area.

THE SAFAH-1x DISCOVERY (1983)

A new concept play evolved out of this regional analysis. The North Lekhwair paleoclosure appeared to be well placed for charge during the Late Cretaceous to early Tertiary. This timing was comparable with that observed in several Lower Cretaceous fields of central and eastern onshore Abu Dhabi. It also seemed possible that the Upper Shu’aiba shelf break might have provided a critical updip seal, capable of preserving a structurally trapped paleo-accumulation from spilling to the south during later Miocene tilting (Figure 7). Nevertheless, the prospect could only be constrained in a very general fashion and selecting the optimum location for an exploratory well to test the concept, proved to be a pragmatic compromise between the most favorable structural location and the larger, more speculative stratigraphic closure. Even though it was generally agreed that the best position for the test well was within the paleoclosure immediately north of the Upper Shu’aiba shelf break, the precise location of this updip seal was poorly defined, with distinct possibility of drilling outside the stratigraphic trap limits. However, several small areas of structural flattening (Figure 5) had been highlighted with the low resolution 2-D seismic available at the time. These at least were comparatively well defined, ranging from 1,000 to 2,000 acres, with an optimistic 10 ms of possible vertical relief.

One of the most robust areas of structural flattening was selected as the location for Safah-1x, some distance to the north of the predicted shelf edge and carefully placed to test both the relatively low- risk structural flattening and constrain the more speculative stratigraphic play concept. A deterministic ‘most likely’ (mode) oil-in-place estimate of 34 MMbbl was assigned to the structural anomaly with perhaps 6 to 8 MMbbl recoverable reserves (Figure 4). Oil-in-place estimates for the stratigraphic prospect ranged widely from 10 million to over 1 billion barrels (P50 = 411 MMbbl). This reserve size envelope was extremely broad—so broad as to have little discriminatory value in comparisons with other competing high-risk ventures. However, at the time the decision was taken to drill Safah-1x, the value of information expected from the well was considered enough to justify its cost.

Safah-1x was spudded on January 2, 1983. The well just tagged the northern end of what was to become the Safah field, intersecting a 14-ft pay section of high-porosity, low-permeability chalky limestone at the top of the Upper Shu’aiba immediately beneath the Nahr Umr Shales. Several tests were attempted. The last of these flowed at 50 to 300 bopd of 37.8° API oil with a gas-oil (GOR) ratio of 1,488 standard cubic feet of gas/barrel (Figure 8)

APPRAISAL (1983 to 1987)

During the latter part of 1983, the Safah discovery was confirmed by two additional wells drilled nearby which tested oil at marginal economic rates. While they clearly demonstrated the existence of an effective updip trapping mechanism, their close proximity and small oil columns provided little insight into the possible size of the accumulation. Furthermore, their poor reservoir quality was discouraging and focused attention upon the still unproven possibility of a Bu Hasa-like high-porosity, high-permeability fairway presumed to exist somewhere to the south.

Trap and Reservoir Architecture

Two alternative trapping possibilities were initially considered: an updip porosity pinch-out within a small isolated shoal or a linear south-facing shelf-margin break as envisaged in exploration model.

  • Shoal Model. Isolated intrabasinal shoal carbonates had been recognized by Hassan et al. (1975), Murris (1980) and Alsharhan (1983) at Jarn Yaphour, Mandous, Zerkouh and elsewhere (Figure 6). Bauman (1983) compared a possibly analogous Upper Shu’aiba reservoir at Al Huwaisah to the modern Bu Tini shoal in the Arabian Gulf (Purser, 1973). In-house interpretations of the Upper Shu’aiba rudist-bearing section at Wadi Rafash (Frost et al., 1983), suggested a broadly comparable model. However, there were significant differences. Most of these locally developed shoal sequences appeared to have developed on salt-cored bathymetric highs, often with algal boundstone precursors. In contrast, the porous Upper Shu’aiba interval encountered at Safah-1x, -2, and -3 was dominated by mud-rich shelf sediments and appeared to overlie flat-lying basinal limestones with little obvious suggestion of pre-existing bathymetric control.

  • Shelf Margin Model. The apparent absence of an algal boundstone precursor and the dominance of low-turbulence carbonate facies, initially suggested that the Bu Hasa shelf margin analog might also be inappropriate. However, the cleaning-up profiles observed in the first three wells provided strong evidence of lateral progradation and supported a low-energy shelf model with the possibility of an older and still undefined algal-platform sequence farther to the northeast (Figure 9). The earlier regional facies interpretation (Figure 6) also had to be readjusted to accommodate new information from the Fahia and Maqtua wells (Figure 9). The Fahia section was re-interpreted as an isolated intrabasinal shoal based on work by Alsharhan (1983) while the Maqtua well was included in the larger platform sequence. Together they imposed a more northerly orientation on the postulated shelf-to-basin transition (inset Figure 9).

Both models (Figure 9) were defended enthusiastically. As they predicted different trapping geometries with quite different potential reserves, there was considerable discussion about the optimum appraisal strategy. The quality of the seismic data was far too poor to define the limits of the field and it quickly became apparent that this could only be achieved by further drilling. Two strategies were considered— a cautious approach stepping out a short way, or a bold jump to the south. Although much safer, a short step-out was considered unlikely to provide any significant new stratigraphic information. In contrast, it was felt that a long step-out well would more effectively constrain future reserve estimates and provide stratigraphic information crucial in locating the shelf-margin porosity fairway suggested by the Bu Hasa analog.

The second approach was favored and Safah-4 was drilled 6 km to the south. It encountered oil with an oil-water contact at about 6,090 ft, consistent with that in the first three wells and confirmed the existence of an areally extensive accumulation. However, much of the Upper Shu’aiba interval was dominated by an impermeable coralgal reef facies, quite unlike that seen to the north (Figure 9). Rather than constraining the reservoir architecture as hoped, the increased complexity implied by Safah-4 suggested that a far larger number of appraisal wells than expected, might be required to adequately define the size of the accumulation.

Although disagreement about the two competing models continued, the results of the Safah-4 well favored the shelf-margin alternative and this was used as a predictive framework in subsequent appraisal planning and well placement. Attention was focused on locating the high-porosity/permeability shelf-margin fairway suggested by the Bu Hasa analog. Ten more wells were drilled in widely spaced locations over the next two years to better constrain the rapidly evolving geological model. These successfully defined the areal limits of the accumulation. Several were cored extensively to provide lithofacies and reservoir data, used to calibrate uncored sections. However, rather than resolving the earlier uncertainty, this additional information only served to highlight the stratigraphic complexity of the reservoir. The porous high-energy shelf-margin fairway predicted by the exploration model was clearly not developed. Instead, the entire field appeared to be dominated by pervasively bioturbated carbonate mudstones, wackestones, packstones, locally developed muddy grainstones and boundstones, with mud-loving requinid rudists rather than the caprinids more typical of Bu Hasa.

Comparison with analogous recent environments provided a possible explanation. Earlier work by Purser and Evans (1973) in the Arabian Gulf had emphasized the importance of wind direction on modern carbonate depositional facies. Wave-dominated linear shorelines exposed to the prevailing Shamal winds in the eastern part of the United Arab Emirates, pass laterally into the tidal coastal complex of central Abu Dhabi and the lower energy mud-rich environments in the west where the coast is sheltered by the Qatar Peninsula. With this analogy in mind, it was argued that the contrasting facies encountered at Bu Hasa and Safah might reflect the prevailing Aptian wind direction from the Neo-Tethyan Ocean in the northeast. Whereas Bu Hasa faced the open waters of the Bab Basin, Safah would have lain on the lee side of this inland sea perhaps in a more protected, less turbulent depositional environment dominated by mud-rich facies (Figure 6).

Several more southerly wells (Safah-7, -8, -9, -11 and -14) encountered distinct cleaning-up calcareous shale-limestone cycles with a deep inner-neritic to shallow inner-neritic fauna. These shales are not present in the north where the Upper Shu’aiba was found to be dominated by massive neritic carbonates with irregular impermeable layers (Safah-1, -2, -3, -10, -12 and -13). In the absence of any distinctive marker horizons, correlation between these two areas was uncertain and so consensus on a most likely stratigraphic model remained elusive (Figure 10).

  • Mid-Shelf Ramp Model. The earlier isolated-shoal alternative evolved into a mid-shelf, sub-fair weather wave base ramp model, dominated by quiet water deposition interrupted by occasional storm events with a slightly deeper water shale and argillaceous lime mudstone facies to the southwest (inset Figure 10). Most significantly, it was argued that in the apparent absence of any definitive intertidal structures or meteoric dissolution features, the entire sequence must have remained submarine throughout and had probably never been subaerially exposed for an extended period of time. As a consequence, its internal architecture was interpreted as a series of weakly organized but semicontinuous, subhorizontal layers.

  • Shelf Margin Model: The differentiated shelf to basin model (Figure 10) was also refined to accommodate the new information. As suggested by earlier interpretations, the Lower Shu’aiba A Member was viewed as a regionally extensive carbonate shelf facies that passed up into a low-gradient aggrading ramp during the mid-Shu’aiba. As sea level rose, this evolved into a differentiated shelf-margin to basin sequence with about 130 ft of topographic relief, prograding basinward toward the southwest. In contrast to the mid-shelf alternative, the influx of calcareous clays and argillaceous limestones during the later Shu’aiba were thought to reflect a sea-level fall with subaerial exposure of the northeastern shelf and restricted lowstand deposition to the southwest. A subsequent somewhat oscillating sea level was implied by the cyclic alternation of shales, limestones and small coralgal patch reefs eventually rising to drown the exposed shelf.

While both models were founded on the same basic observations, their emphasis was different. The first highlighted the generally fine-grained mud-rich character of the sequence, suggestive of quiet, deeper water deposition. In contrast, the second possibility stressed the importance of coralgal boundstones and coated grain facies, locally developed irregular to brecciated bed contacts and mottled textures with small vugs, all widely developed on the northern platform. Although certainly not definitive, it was argued that these features were indicative of low turbulence, shallow-water depositional environments subject to intermittent subaerial exposure.

This slight difference of interpretative focus implied a very different stratigraphic architecture. Instead of a layered reservoir suggested by the mid-shelf ramp scenario, the alternative predicted a profound stratigraphic boundary associated with a linear shelf margin, separating an area of broad reservoir continuity to the north from somewhat more discontinuous, irregularly developed porosity to the south. This found some support in several observations, including an isolated water-wet porous zone far updip in Safah-11, pressure discontinuity between Safah-7 and 8 and a downdip gas cap in Safah-10, farther to the north (Figure 10). It also provided an explanation for the field’s updip seal, quite different from the sharp shelf break predicted by the exploration model, within a sequence of partly to completely unconnected foreshelf to intrabasinal shoal limestones.

Reservoir Quality

By the time the 14 appraisal wells had been completed, it was clear that the Upper Shu’aiba at Safah formed a rather unimpressive, low-productivity reservoir. Test rates averaged only 300 to 500 bopd/well across the field, from as low as 50 to 100 bopd in some southern wells to locally as much as 1,000 bopd or more in the north.

Petrographic analysis combined with log and test data quickly established the reason for this poor performance. It was discovered that virtually all the original macro-porosity within the more grain-rich facies had been occluded by early calcite cement, leaving the muddy matrix of the mudstones, wackestones, and packstones as the primary reservoir rock in the field. Although the mud matrix permeabilities were very low (average 3-5 mD), matrix porosities commonly ranged between 12 to 30 percent (average 22%) and appeared to have been preserved from burial-related compaction by a rigid protective framework of scattered skeletal grains and local calcite cementation. Scanning electron microscopy revealed that these chalky muds were composed of loosely interlocking microrhombic calcite crystals about 3 to 7 micrometers in size. The rhombic crystals were etched and corroded with enlarged pore sizes ranging between 1 and 15 micrometers. The best microintercrystalline porosities appeared to be developed in clean wackestones and packstones of the northern platform. Farther south, the clays in the more argillaceous-dominated facies had clearly encouraged pervasive stylolitization and cementation in the interbedded limestones, with correspondingly poorer reservoir quality.

The origin of these chalky micritic textures has been controversial with several competing hypotheses. Twombley and Scott (1975) attributed lime-mud matrix porosities at Bu Hasa to early dissolution and leaching. Frost et al. (1983) and Harris et al. (1984) subsequently identified an extended late Aptian stratigraphic hiatus of 1 to 1.5 My duration at the Shu’aiba-Nahr Umr contact. They argued that the chalky textures might have developed at that time as a result of widespread meteoric dissolution. Support came from Baumann (1983) and Litsey et al. (1983) who both described meteoric dissolution features in the Al Huwaisah and Yibal fields of north-central Oman and by Alsharhan (1985 a,b) at Bu Hasa and elsewhere.

However, the evidence for subaerial exposure at Safah appeared less obvious. Although some subtle textural and sedimentological evidence of exposure had been observed in a few cores, there was significant disagreement about its field-wide importance. Carbon and oxygen isotopic analysis of the crystalline mudstones provided little support, lacking the depleted isotopic character typical of meteorically altered Pleistocene and Holocene limestone analogs (Figure 11). Instead, it was argued that the Shu’aiba reservoir must have remained submerged throughout deposition and its chalky texture was interpreted as the result of in situ recrystalization at modestly elevated temperatures (48o–53oC) in a closed subsurface system. The corroded crystal fabric was thought to reflect a later, very pervasive dissolution event. Certainly there was little doubt that this had been critical in enlarging the microintercrystalline pore spaces sufficiently to allow the modest flow rates observed in the field. Well behavior suggested a dual or variable permeability reservoir system. Well rates were typically more hyperbolic than exponential in character, leveling off at some lower but stable rate. This was generally attributed to variations in the intensity of microintercrystalline dissolution, with flow into the wellbore limited to more permeable intervals characterized by enhanced dissolution porosity and supported by modest cross-flow from lower permeability matrix rock.

Hydrocarbon Charge

Crude oils from the early Safah wells were napthenic-aromatic in character, of moderate to high maturity (35o–39oAPI and greater) and with a low sulfur content (0.32%). Early geochemical analyses suggested that they originated from a marine carbonate source rock approaching expulsion maturities of 1 percent vitrinite reflectance (VRo). Crudes from wells drilled farther south at a later date were rather more paraffinic, moderately mature (34.8o–41oAPI, VRo = 0.8%) and isotopically more negative. By the time the appraisal program had been completed, a systematic change in crude oil stable carbon isotope character was apparent across the field, ranging from δ13C 26.8‰ in the north to δ13C 30.08‰ to the south. High-resolution gas chromatography revealed a parallel change in the composition of the crude oil with three discrete groups of oils in the north, central and southeast. These compositional and isotopic gradients cut across the more prominent stratigraphic trends of the field, showing very little correlation with its reservoir architecture. Various explanations were proposed for the apparently anomalous relationship including fractionation during migration, and multiple geochemically distinct sources with limited inreservoir mixing. However, nothing was resolved and it remained an enigma fo some time.

The source of the oils was also unclear. More recent oil-to-oil and oil-to-source correlations in the United Arab Emirates (ADNOC, 1984) suggested that the Bab facies highlighted by Murris (1980) and originally assumed to be the most likely candidate for Safah crudes, was only locally developed and of poor to moderate quality at best. However, the broad geochemical similarity of oil in the Arab and Thamama reservoirs throughout Abu Dhabi implied a common origin. Those compositional differences that did occur appeared to be related to differences in maturity, whereas the similarity of their gas chromatography-mass spectrometry and sulfur profile traces, together with a rather uniform carbon isotope character (δ13C 25.3–26.9‰) were considered enough to type the oils to the organically rich radioactive shales of the Hanifa/Diyab intraplatform basin (ADNOC, 1984; Hassan and Azer, 1985; Alsharhan and Kendall, 1986; Alsharhan, 1985a).

These several published syntheses (ADNOC, 1984; Loutfi and El Bishlawy, 1986) also confirmed the earlier interpretation of Murris (1980) which limited the organically richest Hanifa/Diyab source facies to the western offshore and adjacent onshore of Abu Dhabi (Figure 12). Time-equivalent rocks in central and eastern Abu Dhabi seemed to be of only marginal source quality and so an unlikely source for Thamama reservoir oils in that area. However, it soon became apparent that there was a close correlation between deteriorating source quality and increasing maturity in central onshore Abu Dhabi, suggesting the measured total organic carbon values might be largely residual in nature. This was supported by the improvement in source quality observed in the direction of the shallower and less-mature southeastern part of Abu Dhabi and the Lekhwair High (Loutfi and El Bishlawy, 1986). Consequently, it was felt that the Hanifa-Diyab generative area may originally have extended farther east across Abu Dhabi to the border with Oman and might have been well-positioned to charge the Safah field (Figure 12).

Maturity modeling of this exhausted generative area suggested that peak expulsion occurred during the Late Cretaceous moving southeastward during the early Tertiary. The timing appeared to fit well with the charge history at Bu Hasa (Twombley and Scott, 1975) and perhaps explained the relatively late charge seen at Safah with more severe in-reservoir diagenesis. However, it did not satisfactorily account for the isotopic and compositional gradients observed in the Safah crudes. Parallel compositional differences in the maturing source were invoked to account for these differences but the isotopic range was far greater than any observed in presumably equivalent Thamama reservoir oils nearby.

Reserve Estimates

By the end of the appraisal period, oil-in-place estimates had narrowed from a very broad envelope to two quite different alternatives (Figure 4):

  1. Those based on the layered mid-shelf ramp model and a conservative 10,000-acres closure limited to the area around the more successful appraisal wells. Hydrocarbon-in-place estimates ranged from 130 MMbbl to 203 MMbbl oil with approximately 130 billion cubic feet (BCF) of gas.

  2. An optimistic 1,080 MMbbl stock-tank oil originally in place assessment based on the differentiated shelf-to-basin model with a productive area of closure (20,000 acres and 60 ft net pay average) projected outward more boldly from the same scattered well control.

Both alternatives were defended enthusiastically but well rates ultimately proved far more critical in determining the viability of the project. The initial exponential decline rates were discomforting and there was serious concern that the wells might not pay out. This was only mitigated with time as they leveled off at low, but still marginally economic rates.

EARLY DEVELOPMENT (1987 to 1992)

By the time the appraisal program had been completed, the marginal quality of the reservoir had been confirmed with little expectation of high-productivity wells and the economic viability of the field was very much in doubt. The oil price collapse in 1986 exacerbated this uncertainty (Figure 2). The project survived because of two related factors. Limited production had first started in 1984 using extremely rudimentary surface facilities and careful project management which had kept costs to a minimum (McGann et al., 1996). The savings achieved were enough to justify continued appraisal drilling and convinced the Oman Government that economic development would be possible with some additional financial incentives. It was with their encouragement that it was decided to proceed with an aggressive development-drilling program. This was begun in early 1987 and by 1991, more than 100 vertical wells had been drilled, largely focused toward the better quality carbonate platform reservoir to the northeast. Drilling quickly established the areal extent and gross reservoir volume of the accumulation and oil-in-place estimates did not change greatly thereafter (Figure 4). However, the difficulty in reaching a consensus on the stratigraphic architecture of the field had a profound influence on the static reservoir models devised during this period. This in turn affected concurrent dynamic modeling, history matching, production forecasting and forward development strategies.

Reservoir Architecture

Additional well control combined with seismic modeling, confirmed the initially rather tenuous relationship between top Shu’aiba reflector amplitude strength and Upper Shu’aiba porosity. The transition between weak top Shu’aiba reflector amplitudes in the central and northern part of the field with stronger amplitudes to the south (Figure 13) was quite convincingly correlated to the shelf-to-basin facies transition seen within the field. With this encouragement, additional regional amplitude analysis was attempted outside the field area. The Upper Shu’aiba facies shelf-to-basin transition was tentatively followed around a broad embayment toward the southeast, somewhat north of the position predicted during the appraisal phase (inset Figure 14). Several small, isolated mound-like anomalies with weak reflector amplitudes were identified to the south of the shelf margin. Based upon their similarity with Wadi Rafash-1x and Fahia-1, these were interpreted as isolated carbonate shoals. It also became clear that the Maqtua-1 sequence, previously interpreted as part of the main platform, lay much farther outboard than expected and was enveloped within intraplatform basinal mudstones of the Bab Member.

While this apparent regional continuity favored the differentiated shelf-to-basin model, stratal geometries were clearly far more subtle and gentle than the high-relief aggrading to off-lapping organization of the southern basin margin Bu Hasa analog. The complex internal geometry of the field became increasingly apparent as more wells were drilled and in the absence of definitive regional support, the stratigraphic architecture of the field remained an enigma. With time, a thin linear zone of impermeable argillaceous limestones, informally referred to as the ‘shale ditch’, was recognized, dividing the reservoir into the colloquially termed ‘East Lobe’ of massive carbonates and a rather narrower ‘Western Lobe’ that deteriorated rapidly in quality toward the southwest (Figure 14).

A detailed sedimentological review by Kitson et al. (1986), Scott and Kitson (1987), and Scott (1987) refined and amplified earlier observations. They identified 12 distinct sedimentary facies ranging from:

  • outer-shelf calcareous shales,

  • argillaceous and non-argillaceous bioturbated and serpulid-rich shelf limestones to,

  • subtidal peloidal, orbitolinid and oncolitic wackestones and packstones and,

  • rudist-rich intershoal and shoal packstones and grainstones and coralgal biostromes.

The additional control confirmed that the northeastern shelf facies in both lobes was dominated by wackestones and packstones with locally developed coralgal biostromes typical of quiet-water environments. Calcareous shales and argillaceous limestones were clearly restricted to the southwest, alternating with cleaning-up bioturbated limestones. The review also demonstrated that whereas rudist facies were well developed at Safah-9 in the transitional area between the northeastern and southwestern facies associations, they only occurred sporadically elsewhere. As observed earlier, the rudists themselves were dominated by requinids, typically more characteristic of quiet-water environments than the turbulence-loving caprotinids and caprinids found at Bu Hasa (Hamdan and Alsharhan, 1991). However, while there was general agreement upon the quiet-water character of the depositional facies, there was still little consensus on the reservoir architecture of the field, with alternative models predicting either a relatively continuous layered sequence or one divided by a series of gently inclined flow baffles and barriers.

The internal stratigraphic complexity was also highlighted by in-field fluid and pressure discontinuities. This became increasingly apparent over time. Earlier testing and log analysis had established an oil-water contact at 6,110 ft subsea in the central and northern part of the East Lobe. However, subsequent drilling toward the southeast found oil down to about 6,130 ft subsea and eventually a rather variable oil-water contact was recognized that ranged from -6,130 to -6,115 ft in the northwest to -6,120 to -6,130 ft in the southeast. The contact in the West Lobe was even more irregular, although always higher than to the east. Definition of the oil-water contact was complicated even further by locally extensive residual oil columns.

Some degree of reservoir separation also appeared to be implied by the contrasting PVT properties of the oils in the Safah field. During the early stages of development, the entire East Lobe had seemed to be charged by a fairly homogeneous, high GOR oil. As drilling progressed, an ill-defined boundary was discovered (Figures 14 and 15) separating high GOR oils (41°to 54° API, 950 scf/bbl) and associated free gas caps in the northwest, from under-saturated low GOR oils (<40° API, 580 scf/bbl) to the southeast (Vadgama et al., 1991). Further drilling was unable to identify any obvious stratigraphic barrier or baffle that satisfactorily accounted for the contrast.

Serious reservations also emerged about the significance of the linear mid-field ‘shale ditch’ facies tract that separated the East and West Lobes. While the downdip structural position of the gas caps suggested it formed a significant reservoir barrier, this was contradicted by independent geochemical and physical gradients observed in the reservoired crude. Low GOR oils very similar to those in the southeastern part of the East Lobe were present in the immediately adjacent West Lobe. In addition, the three distinct oil families recognized earlier, appeared to cross unaffected from East to West lobes (Figure 15). Both of these observations implied a far greater degree of reservoir continuity than suggested by the stratigraphic interpretation (Figure 14).

Because of the reservoir’s complexity and the difficulty in establishing a working model that resolved these several apparently conflicting observations, it was ultimately decided to abandon the earlier stratigraphic emphasis and use a simpler geometric approach (Vadgama et al., 1991). It was felt that the reservoir architecture could be described quite adequately without imposing any stratigraphic bias, by using an internally consistent correlation grid developed from the increasingly close well control. In subsequent modeling studies, the reservoir was divided into several field-wide flow units alternating with subhorizontal baffles and barriers. These layers were defined entirely by log character, intentionally discounting any preconceived or partly constrained stratigraphic assumptions, while reservoir discontinuities previously attributed to facies changes were re-interpreted as fault barriers (Figure 15).

Reservoir Quality

In an attempt to better define porosity and permeability distribution within the Upper Shu’aiba reservoir, Scott (1987), Scott and Kitson (1987), and Scott et al. (1988) examined its postdepositional history in detail during this period. They recognized three broad stages of lithification and porosity development as follows:

  1. Synsedimentary diagenesis with local marine micritization, marine cementation, and internal sedimentation.

  2. Early burial diagenesis. Contradicting earlier observations, they suggested that the meteoric leaching of aragonitic skeletal detritus had created much of the biomoldic porosity observed in bioclastic packstones/grainstones. Drusy calcite fringe cements and syntaxial echinoderm overgrowths were attributed to freshwater phreatic diagenesis. This appeared to have been followed by a phase of non-ferroan calcite cementation which had occluded most of the intergranular and biomoldic pore spaces and reduced porosity and permeability very severely. As in earlier interpretations, they stressed the importance of the early development of a rigid matrix framework in preserving mud-matrix porosity.

  3. Late burial diagenesis. The increasing importance of chemical compaction with depth was emphasized, particularly in the more argillaceous intervals where dispersed clays appear to have acted as a catalyst for extensive stylolitization with ferroan drusy and blocky calcite cements. As previously highlighted, a second period of relatively late leaching was recognized with extensive dissolution of both lime mud and micritic grains, enhancing the early micro-intercrystalline pore system. A minor phase of late saddle dolomite cementation and replacement was also identified, immediately preceding the migration of hydrocarbons into the reservoir.

Although Scott et al. (1988) had confirmed the importance of intercrystalline microdissolution porosity in the Safah reservoir, its origin and distribution remained controversial and both early meteoric and deep subsurface hypotheses continuing to find support. This uncertainty had important practical implications in ongoing reservoir modeling, both in predicting permeability distribution and connectivity and also more indirectly in understanding the stratigraphic architecture of the field.

In an attempt to resolve this issue, Shu’aiba diagenesis was reviewed on a wider stage. Moshier et al. (1988) and Moshier (1989b) had proposed a model of ‘arrested cementation’ based on their work in the Sajaa field to the north. As with the chalky micrites at Safah, the Sajaa reservoir was dominated by high micro-intercrystalline matrix porosities, apparently quite unrelated to meteoric diagenesis. As at Safah, isotopic analysis (Figure 11) of the matrix carbonates revealed a slight negative δ18O shift and relatively unchanged δ13C compared with values estimated for original Aptian carbonate sediments (Scholle and Arthur, 1980; Moldavanyi and Lohman, 1984; and Moshier, 1989a). Moshier et al. (1988, 1989b) argued that this was consistent with precipitation in fluids of marine composition within a relatively closed system and suggested a process of in situ dissolution and reprecipitation at shallow depths (about 380–550 m) and low temperatures (36°–41°C) during the stabilization of lime muds. As this process involved little mass transfer, they suggested the high matrix porosities had been inherited from the original muds and largely preserved from later mechanical or chemical compaction by the interlocking microrhombic fabric. A late phase of microcrystalline corrosion was also recognized, but this was considered to be of less significance.

The Moshier (1989b) model was rejected by Budd (1989) as a result of his studies of comparable microcrystalline Shu’aiba porosity in the nearby Margham field. He suggested a two-stage process of formation within a meteoric environment. This was founded upon the extremely low Sr and Mg content of the microcrystalline calcites which he attributed to an initial phase of extensive flushing in an open, water-buffered aquifer immediately after Shu’aiba deposition. He argued that subtle δ18O isotopic differences in petrographically identical high- and low-porosity microcrystalline limestones must reflect primary porosity-controlled variations of rock-to-water buffering during later recrystalization within a confined meteoric aquifer. Although this model was supported by regional stratigraphic evidence of uplift and exposure that followed Shu’aiba deposition, it remained difficult to explain the undepleted δ13C nature of the micrites, apparently unaffected by what must have been δ12C-enriched meteoric waters.

These published studies provided a critical conceptual framework to refine and extend later petrographic analyses of the Safah reservoir. Following initial work by Prezbindowski and Orban (1989), Prezbindowski et al. (1990) presented a new subsurface diagenetic/structural explanation for micro-intercrystalline porosity development in the Safah field. While they recognized the significance of early mineral stabilization and development of microrhombic crystalline fabrics, they discounted its importance in the development of reservoir-grade permeability at Safah. Instead, they argued that this had been created by a late dissolution event post-dating early stylolitization and proposed the following diagenetic sequence:

  1. Initial mechanical compaction, carbonate mineral stabilization and recrystalization during early burial as described by Moshier (1989b).

  2. Equant spar calcite cementation occluding all intra- and inter-particle primary porosity.

  3. Early stylolitization with an initial phase of microfracturing in response to tectonic readjustment or unloading.

  4. Spar-calcite cementation of the first phase microfractures.

  5. A second generation of open microfracturing associated with unroofing or structural movement.

  6. Development of microdissolution and grain-moldic porosity by the preferential etching and solution of very fine crystalline carbonate components including peloids, benthic forams, mud-matrix material and individual rhombic crystals. Whereas the second-generation microfractures were considered critical in allowing corrosive fluids to permeate through the reservoir, dissolution porosities appeared best developed in limestone facies with a significant component of skeletal grains. It was argued that these had selectively preserved higher permeabilities both by shielding the mud matrix from later compaction and by encouraging local differential fracturing. This in turn would have preferentially focused later fluid through-flow and resulted in more aggressive dissolution.

  7. Hydrocarbon entry into the reservoir. Systematic, diagenetically related changes in reservoir quality across the oil-water contact like those observed in fields nearby (Harris et al., 1968; Johnson and Budd, 1975; Litsey et al., 1983; and Burgess and Peter, 1985) were difficult to establish. However, the reservoir interval had clearly undergone significant diagenesis, suggesting that it had been charged relatively late in its history.

The two periods of microfracturing were tentatively correlated to periods of regional structural readjustment during the Late Cretaceous and Tertiary (Glennie et al., 1974; Patton and O’Connor, 1988; Warburton et al., 1990). Burruss et al. (1985) had previously described Late Cretaceous microfracturing events in Lower- mid-Cretaceous carbonates of the nearby foredeep, which provided some independent support for this hypothesis. Calcite cements infilling these fracture systems were commonly associated with hydrocarbon and aqueous inclusions. These testified to at least two periods of petroleum expulsion and migration, which might have generated the corrosive organic acids and CO2- enriched fluids responsible for the late-stage dissolution event observed at Safah on the North Lekhwair Arch (Prezbindowski and Orban, 1989; Prezbindowski et al., 1990; and Prezbindowski, 1991).

This diagenetic model provided a useful conceptual framework for predicting the dominant pore distribution throughout the field (Prezbindowski and Orban, 1989; Prezbindowski, 1991):

  1. Better quality reservoir-grade permeability was restricted to the matrix of clean wackestones and packstones. It was only well developed in mudstones where it was directly associated with microfractures.

  2. Dissolution porosities in limestones associated with clay-rich intervals had suffered from more severe stylolitic-related cementation.

  3. Vertical to horizontal permeability ratios approached unity in cleaner, massively bedded carbonates, but varied locally in areas of more intense microfracturing and in intervals with interbedded clay-rich layers.

  4. Clay and argillaceous limestone reservoir barriers and baffles were amplified by more extensive stylolitization.

  5. Dissolution intensity appeared to increase upward through the reservoir section. This apparently reflected a parallel increase in the density of microfractures toward the top of the Shu’aiba, probably related to the brittle-ductile mechanical boundary contact with the Nahr Umr Shales.

Most significantly, the analysis suggested a subtle causal link between carbonate facies and reservoir quality. However, this was of only limited value initially because of the difficulty in defining the field’s stratigraphic architecture and reservoir descriptions that at the time, relied on permeability distributions derived from stochastic log-based estimates. Subsequent history matching and reservoir simulations highlighted the limitations of this approach and more emphasis was placed on stratigraphic controls later in the life of the field.

Hydrocarbon Charge

As development continued, further analysis of Safah crudes refined earlier interpretations and appeared to confirm the presence of three geochemically discrete groups (Figure 15) in the north, central and southeastern parts of the field (Lindberg et al., 1990). However, the significance of this distinction became uncertain when a more rigorous analysis in 1991 suggested that the Safah crude came from a single, mature, mixed terrestrial and marine organic source (Geochem, 1991). This disconcerting interpretation was based upon whole-oil chromatography that gave consistent ratios of pristane:phytane (1.16–1.35), pristane:nC7 (0.23–0.33), nC7:c7 napthene (0.7–0.93), and total methylcyclohexane: dimethylcyclopentane (1.92–2.39) throughout the field. In addition, the sulfur content was universally low (0.2–0.8%) and calculated expulsion maturities fell within a narrow range of 0.74 to 0.84 percent VRo. Because of this uniformity, the more subtle compositional and isotopic gradients observed previously were considered to be the result of migrational fractionation rather than dual sourcing. It was also argued that the hopane distribution combined with small amounts of moretane and normoretane biomarkers and the absence of oleanane indicated a relatively strong contribution from pre-Maastrichtian higher land plants, suggesting a late Paleozoic to Mesozoic source.

Contemporary reports revealed the ongoing uncertainty about the relative significance of the Bab, Hanifa-Diyab and older source rocks in the region. Alsharhan (1989), Marzouk (1989) and Lijmbach et al. (1992) all acknowledged the possible importance of the Bab source locally. Indeed, Lijmbach et al. (1992) tentatively typed its more carbonate-dominated facies to many of the Thamama reservoired oils in the Ramaitha, Jarn Yaphour, and Sahil fields in central and eastern Abu Dhabi. However, Azer (1989), Alsharhan (1989) and Marzouk (1989) continued to stress the Late Jurassic Hanifa-Diyab as the primary source of both Jurassic and Lower Cretaceous accumulations in that area.

Perhaps the most significant reviews published during this period were by Grantham et al. (1988) and Grantham et al. (1990). They recognized five distinct crude-oil families in Oman and defined their geochemical characteristics in some detail. A comparison of these type oils with those at Safah presented a plausible explanation for the compositional and isotopic variations seen there (Figure 16). The δ13C isotopic character of oils in the East Lekhwair and Sahmah fields farther south in central Oman were comparable with the maximum and minimum values observed at Safah. Grantham et al. (1988) tentatively typed the Sahmah oils to the Sahmah Member of the Silurian Safiq Formation. This was supported by their resemblance to equivalent Qusaiba Shale-sourced oils in central Saudi Arabia (Mahmoud et al., 1992; Abu-Ali et al., 1991). The East Lekhwair oils appeared geochemically identical with oils from the Arab and Thamama reservoirs in the United Arab Emirates to which they attributed a Diyab-Hanifa source. The isotopic similarity of the Safah end members with these two oil families argued for dual charging from basal Silurian and Upper Jurassic source rocks (Figure 16) and contradicted earlier interpretations which had emphasized the otherwise homogenous geochemical character of the reservoired crude.

Reserve Estimates

More than 25,000 acres of productive closure was established by drilling during the early development period and by 1991 hydrocarbons-in-place were assessed at 790 MMbbl oil and 770 BCF gas (Figure 4). Because of the typically poor well performance, recovery factor estimates were very low, ranging between 11 and 15 percent.

LATE DEVELOPMENT (1992 to 2000)

The commercial success of the Safah field encouraged further exploration in the remaining part of the Suneinah Concession. Concepts, patterns and interrelationships first recognized at Safah were used to develop a variety of structural, stratigraphic and combination structural-stratigraphic prospects elsewhere. Starting in 1991, the Occidental-Neste Group embarked upon an aggressive exploratory drilling program. Mkasa-1 (see Figure 19) was drilled to evaluate the crestal part of the pre-Miocene paleohigh to the north of Safah field, but found only residual gas shows. Exploration toward the southeast of the license area proved more rewarding. Al Barakah was discovered in 1991/1992 and Wadi Latham in 1993 followed by several smaller accumulations farther to the east (see Figures 2 and 19). Exploratory success was initially rather checkered because of the difficulty in delineating the more subtle stratigraphic and structural anomalies typical of an area with 2-D seismic data. However, as regional 3-D seismic was acquired, better descriptions of analog prospects were developed and success rates increased rapidly, only to fall again once these had been exhausted.

The dramatic increase in seismic and well data acquired at this time provided much valuable stratigraphic information. In addition to the new exploratory wells, the first 3-D seismic survey was shot in 1994 and was rapidly followed by several more. By 1998, the Safah field area had been covered by single merged 331 sq km 3-D data volume. At the same time, a series of excellent reviews were published providing a sophisticated appreciation of regional Shu’aiba stratigraphy, diagenesis and hydrocarbon environment. This combination of improved regional perspective, more appropriate local analogs and close in-field well and 3-D seismic control acted as a catalyst in refining the geological model of Safah field.

Trap and Reservoir Architecture

Regional Stratigraphic Perspective

During this period, earlier lithofacies descriptions of the Lower and mid-Cretaceous were redefined within a sequence stratigraphic framework (Figure17). The Lower Cretaceous Thamama Group was reinterpreted as a second-order supersequence, aggrading and prograding northeastward across the Arabian continental margin toward the Tethyan Ocean (Connally and Scott, 1985; Haan et al., 1990; Pratt and Smewing, 1990,1993a,b). The Shu’aiba Formation was recognized as the last of several higher order composite cycles within this larger scale sequence and its earlier two-fold ‘A-Member’ ramp and differentiated shelf-basin subdivision was revised significantly.

Although earlier work by Harris et al. (1968), Hassan et al. (1975), Twombley and Scott (1975) and Murris (1980) had highlighted the regional continuity of the Lower Shu’aiba and the broadly transgressive, aggrading character of the Upper Shu’aiba, the detailed stratigraphic architecture and chronostratigraphic relationship between shelf and basinal facies had been poorly understood (Figure 6). However, in 1990 Abou-Choucha and Ennadi published a sequence stratigraphic interpretation of the Bu Hasa field, broadly reminiscent of the differentiated shelf model suggested earlier for Safah. This was subsequently expanded by Calavan et al. (1992) who used high-quality seismic control to divide the Shu’aiba into three, third-order sequences. The upper part of the Hawar Member of the Kharaib Formation and the immediately overlying Lower Shu’aiba formed the first of these, with a basal erosional sequence boundary passing up through transgressive shales into highstand ramp carbonates of the A Member (Azer and Toland, 1993; Boichard et al., 1995). They viewed the second sequence as a composite unit with:

  • A basal transgressive tract represented by an aggrading algal platform and condensed basinal equivalents to the north;

  • An aggradational to progradational early highstand system tract with low-relief shelf to shelf margin sigmoidal clinoform geometries passing northward into condensed basinal equivalents; and

  • a strongly progradational late highstand system with oblique clinoforming geometries, made up of several higher frequency cycles of porous to non-porous rudist-rich carbonates facing north into thin basinal limestones of the intrashelf basin.

They also recognized a third, younger sequence of late Aptian age, onlapping the rudist shoals of the earlier shelf complex from both north and south. To the north of Bu Hasa, this consisted of calcareous shales and argillaceous lime mudstones and basin-margin skeletal-oolitic grainstones, which had previously been considered time-equivalent to the carbonate platform and grouped together with the clay-free intrashelf basinal facies of the underlying sequence. However, new seismic and biostratigraphic evidence allowed Calavan et al. (1992) to draw a critical distinction between the two and redefine the Bab Member to include only the younger, basin-restricted argillaceous package. Similarly, in the southern part of Bu Hasa, better seismic control highlighted the onlapping relationship of a channel-based carbonate unit previously viewed as the lateral shelf-correlative of the rudist shoal complex. This was re-defined as a younger channel-fill to transgressive systems tract, time-equivalent with the basinal Bab Member in the north. The revised interpretation implied that the north-facing rudist shoal complex might have had significant residual relief during deposition of the later Aptian sequence. Shu’aiba subaerial exposure and dissolution had long been recognized at Bu Hasa, and from this refined stratigraphic model, Calavan et al. (1992) were able to suggest that it had been limited to the more elevated parts of the platform and occurred contemporaneously with the deposition of the younger sequence.

Witt and Gökdag (1994) described a similar relationship across the southern margin of the Bab re-entrant in north-central Oman, between the Al Huwaisah and Lekhwair fields (Figure 1). Improved biostratigraphic resolution allowed them to distinguish a time-equivalent north-facing highstand sequence at Al Huwaisah with an algal-mound precursor, passing up into a rudist-rich shoal margin. As at Bu Hasa, this appeared to have been exposed during the mid to late Aptian, restricting later deposition to the basin with an off-lapping sequence of peloid-orbitolinid packstones and wackestones, prograding into argillaceous wackestones-mudstones and occasional re-sedimented packstone/grainstone interbeds to the north. The proximal part of this younger sequence now forms the reservoir at Yibal field.

Later syntheses by Fitchen (1995), Fischer et al. (1995a,b), Azzan and Taher (1995), Taher (1996,1997), van Buchem et al. (2000) and van Buchem et al. (2002) refined and enhanced the Upper Shu’aiba stratigraphic model. Aldabal and Alsharhan (1989) and Azer and Toland (1993) had recognized Bab shale-carbonate cycles earlier in the Zakum field. Fitchen (1995) and Taher (1996,1997) described the Bab Member architecture at Bu Hasa in some detail with a basal onlapping shale member followed by at least three off-lapping carbonate cycles. The ubiquity of this model was supported by Masse et al. (1997) based upon extensive Shu’aiba outcrops in the Jebel Akhdar region of the Oman Mountains where they defined a lower Aptian carbonate platform and platform-margin sequence facing northward toward the Tethyan Ocean (Figure 17). As at Bu Hasa and Al Huwaisah, this appeared to have been exposed during the late Aptian, while the adjacent basin was filled by a shallowing-up carbonate sequence (Al Hassanat Formation), onlapping and eventually flooding back onto the platform during the early Albian. In 2001, a comprehensive sequence stratigraphic synthesis by Sharland et al. (2001) was published supporting and refining this earlier work. In a series of regional paleogeographic reconstructions of the Shu’aiba Formation, they followed the evolution of the southern Arabian mid-Aptian highstand shelf and intraplatform basin sequence through a period of mid- to late-Aptian sea level fall and subaerial exposure of the shelf with contemporaneous deposition of the basin-restricted Bab Member. More support came from a recent analysis of the Kharaib and Shu’aiba of northern Oman and the United Arab Emirates by van Buchem et al. (2002).

Age estimates of the intra-Shu’aiba sequence boundary have varied considerably since Safah was discovered. Initially, a rather approximate late Aptian age was assigned to the contact, but Harris et al. (1984) later tied it to a mid-Aptian global sea-level fall event (110–112 Ma) identified by Vail et al. (1977). Subsequently, attempts were made to correlate it with the Haq et al. (1988) intra-early Aptian LZB3-LZB4 sequence boundary (112 Ma) by Abou-Choucha and Ennadi (1990) and Calavan et al. (1992). Later correlations with the intra-Upper Aptian LZB4.1-LZB4.2 boundary (109.5 Ma) were proposed by Alsharhan and Kendall (1991), Kendall et al. (1991), Simmons et al. (1992), Azzan and Taher (1995), and Fitchen (1995). However, good biostratigraphic control in the basinal sequence between Al Huwaisah and Lekhwair led Witt and Gökdag (1994) to suggest an intra-early Aptian age for the boundary, apparently unrelated to either of the two global events defined by Haq et al. (1988).

More recent recalibration of the Cretaceous time scale by Gradstein et al. (1995) and of the global cycle chart by Hardenbol et al. (1998) necessitated further revision. Sharland et al. (2001) identified an intra-Shu’aiba mid-Aptian (117 Ma) sequence boundary apparently coincident with the Hardenbol et al. (1998) Ap4 (117.07 Ma.) global sequence boundary (modified from the earlier Haq et al LZB 4.1-4.2 event). However, they suggested that the period of maximum intra-Shu’aiba sea level fall and exposure occurred sometime later during the late Aptian (114 Ma) with no equivalent event on the Hardenbol et al. (1998) global cycle chart. Ultimately, attempts to determine the age of this intra-Shu’aiba sequence boundary by correlating it with what may be questionable global eustatic events (Miall and Miall, 2001) appear rather circular and unconvincing. This is particularly so in view of existing chronostratigraphic imprecision and regional tectonic effects (Powers et al., 1966; Murris, 1980; Hughes Clarke, 1988; Kendall et al., 2000). Nevertheless, while the absolute age of this intra-Shu’aiba lowstand sequence remains uncertain, its local synchroneity finds increasing support from recent reports (Masse et al., 1997; van Buchem et al., 2000; Sharland et al., 2001; van Buchem et al., 2002) which emphasize its geometric similarity throughout the southern Arabian Basin. Final confirmation must await the development of a reliable, locally consistent chronostratigraphic framework.

The contact between the Shu’aiba and the overlying Nahr Umr Formation varies significantly in character throughout the region. Although mild erosion and onlap has been observed at the top of the Bab Member of the Shu’aiba Formation, it generally appears to be conformable and continuous into the overlying Nahr Umr (Witt and Gökdag, 1994). The Sharland et al. (2001) reconstruction shows the Bab Basin progressively narrowing through the later Aptian with falling sea level until the Bab-Nahr Umr contact at 114 Ma and then rising once more in the late Aptian in a manner reminiscent of a forced regression. Elsewhere in the United Arab Emirates, the contact between the Shu’aiba platform carbonates and basal Nahr Umr shales was universally described as abrupt but with little evidence of significant erosion (Alsharhan, 1985a, 1987, 1991). However, when traced westward, these shales appear to pass into a diachronous series of transgressive fluvial to shallow-marine sandstones (Soliman and Shamlan, 1982; Alsharhan and Nairn, 1988; Sharief et al., 1989; and Davies et al., 2002) resting upon an increasingly severe erosional unconformity toward southern Oman and central Saudi Arabia (Hughes Clarke, 1988; Alsharhan and Nairn, 1988). Although the contact between the Wasia and Thamama Groups had traditionally been placed at the top of the Bab Member, this larger regional context appears to suggest that the Bab forms the basal lowstand systems tract of the Wasia Group passing up into transgressive Nahr Umr clastics. Consequently, we would argue that the boundary between the Lower and mid-Cretaceous supercycles should be redefined as the contact between the Bab and the older Shu’aiba platform-to-basin sequence (Figure 17), approximately equivalent to the Hardenbol et al. (1998) Ap4 (117.07 Ma) sequence boundary.

Local Stratigraphic Framework

The additional well and seismic control, acquired during the 1991 to 1999 exploratory program, formed the basis for the most significant advance in Shu’aiba stratigraphic interpretations in the Suneinah area since the Safah discovery 10 years earlier. This new information made it possible to divide the Formation into several regionally correlative sequences (Figures 18 and 19). As in earlier reviews, the lower Shu’aiba A Member has been interpreted as a highstand ramp, forming the substrate for a mid-Shu’aiba third-order composite package of three higher frequency cycles:

  1. A transgressive to early highstand Lower Shu’aiba sequence represented by a rather poorly constrained Lithocodium-Bacinella algal-dominated shelf to the east (Frost et al., 1983) with thin basinal equivalents throughout the central and western part of the Suneinah area (mS1). Stacking patterns within enclosing younger units hint at more isolated algal banks north of Mkasa and in the Maqtua area (Figure 18).

  2. A mid-Shu’aiba highstand system (mS2–mS3) represented by a west-facing non-rimmed carbonate platform, composed of skeletal, foraminiferal and peloidal wackestones, packstones, occasional grainstones and muddy caprinid-dominated rudist shoals (Frost et al., 1983) to the east passing westward into basinal mudstones and wackestones (Figure 18). The paleogeography in the central part of the Suneinah area was revised significantly when it became apparent that the shale-carbonate cycles in the Shams wells were very similar to the Bab facies elsewhere in the region and so were probably of late Shu’aiba age. This revision suggested mid-Shu’aiba basinal facies extended northward toward Shams and isolated the Safah-Mkasa shelf carbonates from the eastern platform (Figure 19). Time-equivalent facies in the Safah area have been tentatively divided into two high-frequency sequences (mS2 and mS3). The lower of these (mS2) consists of a laterally continuous series of thin-bedded wackestones and mudstones, aggrading and gently offlapping southward from Mkasa (Figure 18). While entirely speculative, we would argue that the offlapping stratal geometries must have developed by outbuilding from some pre-existing, perhaps algal-cored depositional high, and may in fact represent the leeward fringe facies of a north- or northeast-facing shoal, reminiscent of the modern Bu Tini bank (Purser, 1973). In the Safah area, this homoclinal ramp provided the substrate for a third late-highstand sequence (mS3) of low-relief, low-turbulence carbonate shoals that prograded southward to form a detached non-rimmed open-marine shelf (Figure 19).

    The mid-Shu’aiba composite sequence was terminated by a significant fall in relative sea level. Although there is little obvious evidence of karstification or erosive downcutting, intermittent subaerial exposure is supported by stratal organization, subtle diagenetic fabrics and by correlation with the very similar mid-Aptian event described by Witt and Gökdag (1994) in central Oman, Masse et al. (1997) in the Oman Mountains and by van Buchem et al. (2000,) van Buchem et al. (2002) and Azzan and Taher (1995) in Abu Dhabi.

  3. Upper Shu’aiba sedimentation appears to have been confined to the mid-Shu’aiba intraplatform basin and is interpreted as a lowstand systems tract of a late Aptian to Albian composite sequence. Where observed nearby at Al Barakah and South Salmah (Figure 18), the basin margin facies is typically composed of several offlapping high-frequency cycles (US1-2, 3, 4 and 5+). Each has a basal onlapping basinal facies passing up through wackestones and packstones into grainstone shoals and locally developed rudist biostromes. More basinward equivalents are dominated by shales and thin-bedded argillaceous limestones. Although broadly comparable to time-equivalent sequences described by Calavan et al. (1992), Witt and Gökdag (1994), Azzan and Taher (1995), Fitchen (1995), Taher (1996, 1997) and others, the younger basin-margin cycles at Safah seem to have a more retrogradational and transgressive stratal organization and pass up gradationally into the overlying Nahr Umr Shales (Figure 18).

Current Safah Field Interpretation (2000)

The steady advance in both regional and local understanding of Upper Shu’aiba stratigraphy between 1995 and 1999 proved critical in developing a more sophisticated geological model of the Safah field. This was supplemented by 3-D seismic (Figures 13 and 20) acquired between 1994 and 1997. Although the resolution of the seismic data was still insufficient to reveal much of the internal reservoir geometry, it demonstrated that in-field faulting was far more subdued than previously suggested and earlier fault-based models were subsequently abandoned. Attempts were made to describe the reservoir using equal-thickness slice maps (Cleveland et al., 1996), but although these provided an efficient way of integrating the ever-increasing amounts of detailed lithological, petrophysical and production data, they ultimately proved to be rather limited. Attention once again focused on interpreting the reservoir architecture of the field from a stratigraphic perspective and the earlier differentiated shelf-to-basin model regained favor. This was refined using information and ideas from regional concepts, local analogs and in-field well and seismic control and led to the stratigraphic interpretation illustrated in Figure 21. Although some ambiguity still remained, support provided by nearby subsurface and outcrop analogs was very persuasive.

East Lobe. Earlier facies descriptions were broadened (Harland, 1996) to include information from later wells (highstand vertical sequence model of Figure 22). Three more prominent facies associations were recognized:

  • Burrowed and serpulid-rich outer-shelf carbonate mudstones deposited below the fair-weather wave base. Thin coarse-grained shelly layers containing echinoderms and thin-shelled bivalve detritus and occasional sponge spicules suggested intermittent storm events.

  • Subtidal, midshelf to outer-shelf pelletal and orbitolinid wackestones and packstones characterized by diverse fauna, with occasional oncolitic and coated grain-rich horizons indicating more turbulent conditions.

  • A low-turbulence inner-shelf to shoal facies consisting of requienid-dominated rudist packstones, poorly sorted grainstones and locally developed biostromes of corals, stromatoporoids, and encrusting algae.

Although rather obscured by severe bioturbation, these facies appear to be organized in thin cleaning-up cycles. Closely spaced well control has emphasized the low-relief aggrading to prograding geometry of the East Lobe (Figure 21) and suggested that the cycles stacked up to form the early to late highstand cycle sets of high-frequency sequences mS2 and mS3. Dense, low-porosity intervals at the tops of later progradational cycles (mS3) have been attributed to both submarine cementation and meteoric diagenesis during short-lived exposure.

West Lobe. With the acquisition of 3-D seismic, far more powerful amplitude extraction techniques became possible than with the earlier 2-D data (Figures 13 and 20). The mid-Shu’aiba composite sequence shelf margin was clearly defined by a field-wide change in reflection amplitude of the near top Shu’aiba event. Reflector amplitudes appear uniformly low throughout the East Lobe but change across the shelf margin (‘shale ditch’ or ‘inter-lobe trough’ of earlier interpretations) into a fore-margin zone of more irregular amplitude distribution, grading southward into a region of generally high amplitudes. The change mirrors the development of two distinct cycle sets within the zone of more irregular amplitudes reflecting the develoment of a carbonate-dominated lowstand complex (US1, 2) aggrading and building-out from the mid-Shu’aiba shelf margin (Figure 21). From the limited core control available, this complex appears to consist of:

  • thin bioturbated mudstones and wackestones;

  • pelletal and oncolitic subtidal wackestones and packstones; and

  • well-developed coralgal biostromes with rudist-rich packstones and poorly sorted grainstones.

Unlike the East Lobe mid-Shu’aiba sequence (mS-3), some crude particle sorting and imbrication suggested more agitated conditions and locally developed shoal-top beaches with evidence of subaerial exposure.

Southern Fringe. Farther outboard to the south, higher top Shu’aiba amplitudes (Figure 20) reflect a still younger sequence (lowstand vertical sequence model of Figure 22) consisting of the following units:

  • Fissile and non- burrowed basinal Orbitolina-bearing shales, passing up into weakly laminated and nodular burrowed calcareous mudstones, with thin irregular locally developed intervals of storm-winnowed bioturbated bioclastic wackestones.

  • Bioturbated argillaceous limestones with a relatively sparse marine fauna, grading up into orbitolinid and peloidal wackestones and packstones with more diverse fauna.

  • Locally well developed rudist packstones, grainstones and floatstones, commonly containing abraded requienid and caprotinid rudist fragments.

These facies appear to be organized into a series of high-frequency sequences each with onlapping shales (Figures 18 and 21) and passing up into prograding carbonate-dominated shoals (US 3,4 and 5+). Although still rather controversial, this stratal architecture is supported by comparisons with equivalent better-defined sequences at Al Barakah and South Salmah (Figure 18).

The Upper Shu’aiba Bab-equivalent reservoir sequence at Safah (US1, 2, 3, 4, and 5+) is clearly basin-restricted and younger than the mid-Shu’aiba platform to the northeast. While some ambiguity remains, its stratal architecture suggests deposition during and following a significant base-level fall. The older carbonate-dominated, high-frequency US1 to US2 cycles in this lowstand systems tract, are interpreted to be an attached forced regressive wedge. Onlapping shales within the succeeding US3 sequence may represent the initial advance of the Nahr Umr coastal depositional system into the region from the uplifted western hinterlands. The generally retrogradational character of the later Shu’aiba sequences (US4, 5+) and transition into the overlying Nahr Umr shales, suggests a gradual rise of sea level and passage into the transgressive, backstepping, clastic-dominated sequence set of the Lower Wasia (Figure 21).

There has been considerable recent discussion (Plint and Nummedal, 2000; Posamentier and Morris, 2000) about the relationship between higher order sequence boundaries and forced regressive deposits. Following arguments by Posamentier and Morris (2000), who stressed the importance of defining such boundaries by their more universal character unrelated to differences in local physiography, we favor placing the mid-Aptian event (near Ap4–117.02 Ma supersequence boundary after Hardenbol et al., 1998; Figure 23) at the contact between the mid-Shu’aiba platform (mS3) and the following forced regressive wedge (US1, 2). This interpretation conflicts with that of Abou-Choucha and Ennadi (1990), Calavan et al. (1992), Witt and Gökdag (1994), Fitchen (1995), and Taher (1996, 1997) who defined the event at the first appearance of onlapping calcareous shales. More recently, van Buchem et al. (2002) described equivalent forced-regressive deposits to the south and west of the Safah field, and placed them in the preceding late highstand sequence. The issue remains to be resolved.

This more recent interpretation of Safah’s reservoir architecture predicts several orders of internal stratigraphic baffles and barriers to reservoir fluid flow. These include:

  • laterally continuous barriers representing the deeper water facies of the aggrading (mS2) high-frequency sequence of the East Lobe;

  • dense, cemented baffles at the tops of the prograding high-frequency sequence cycles (mS3) in both the East and West lobes;

  • those associated with the second-order sequence boundary between the mid and Upper Shu’aiba;

  • the boundary between the forced regressive carbonate wedge and the younger more argillaceous dominated lowstand sequences of the Upper Shu’aiba; and

  • calcareous shales and associated stylolitic seams that separate the lowstand shoals.

The still only partially homogenized character of the dual-sourced oil suggested that these stratigraphic barriers and baffles might have significantly hampered in-reservoir fluid flow and mixing. Current fluid sampling frequency is insufficient to detect any more subtle compositional changes that might reveal their relative importance. However, they are expected to have a material impact upon the ongoing waterflood program and may have an important effect upon the productive life of the field.

Reservoir Quality

Regional Perspective

During the later phases of field development, new regional analogs, information, core and seismic data were combined with the re-examination of old material (Harland, 1996) to give fresh insights into the critical factors controlling permeability in the Shu’aiba reservoir at Safah. These included:

  • Upper Shu’aiba Subaerial Exposure. As more wells were cored in the Safah field, further signs of syndepositional and early postdepositional meteoric diagenesis began to emerge. Discrete and diffuse fissures were recognized in several intervals, extending downward from sharply defined exposure surfaces into more diffuse thoroughfares below (Figure 22f). Fissures at the top of the Shu’aiba reservoir were filled with shale and pyrite from the overlying Nahr Umr Formation, whereas those within the reservoir sequence contained skeletal detritus, small matrix clasts and even weathered rudist fragments. Vadose fitted textures, oversized solution-enlarged interparticle pores, elongated solution vugs and geopetal fills were also identified locally and point to intermittent periods of subaerial exposure (Harland, 1996).

    Over time, evidence of Upper Shu’aiba subaerial exposure began to be identified elsewhere in the region. A 1-to 2-My biostratigraphic hiatus had been recognized for some time at the outcropping Shu’aiba-Nahr Umr contact in the Oman Mountains (Simmons and Hart, 1987; Scott, 1990; Simmons, 1994). Nevertheless, little or no definitive physical or chemical support of meteoric diagenesis (Wagner, 1990) could be found. Evidence of top Shu’aiba subaerial exposure remained equivocal (Wagner, 1990; Pratt and Smewing, 1993a,b; Russell et al., 1996) until Immenhauser et al. (1999) and Immenhauser et al. (2000) discovered significant isotopic excursions and brackish-water fluid inclusions in the same area. The apparent absence of caliche and karst features at the contact had previously seemed to discount the possibility of significant exposure (Wagner 1990). However, Immenhauser et al. (1999) and Immenhauser et al. (2000) argued that karstification may have been subdued by the extremely low relief of the Shu’aiba platform and explained the apparent absence of paleosols at the contact by erosional reworking into succeeding transgressive deposits.

    Farther to the east, a prolonged period of top Shu’aiba exposure had been demonstrated from petrographic and textural evidence at Al Huwaisah, Qarn Alam, Ghaba North, Shaybah, and Bu Hasa (Witt and Gökdag, 1994; Vahrenkamp and Grötsch, 1995; Vahrenkamp, 1996; Al-Awar and Humphrey, 2000; Hughes, 2000; and Ziegler, 2001). Somewhat disconcertingly, none of the carbonates described immediately underlying the contact, showed the strongly negative δ13C isotopic character, previously considered a defining characteristic of subaerial exposure and fresh-water diagenesis (Figure 11). Various explanations were put forward to account for this. Al-Awar and Humphrey (2000) suggested that meteorically controlled depletion might have been severely attenuated in a dry, arid climate with limited vegetation and δ13C-enriched soils. However, this found little support in Aptian climate reconstructions (Barron and Washington, 1985; Francis and Frakes, 1993), which tended to favor more humid conditions. Although this issue has still to be resolved, the more comprehensive regional perspective developed during this period discounted the earlier isotope-based arguments against meteoric diagenesis at Safah and supported the still rather ambiguous petrographic evidence of field-wide subaerial exposure. Nevertheless, the precise origin of both matrix micro-intercrystalline fabrics and subsequent dissolution porosity remains controversial.

  • Early Mineral Stabilization. There is now general agreement that the ubiquitous microrhombic crystalline fabric of the Safah reservoir limestone developed shortly after deposition as a result of in situ mineral stabilization. As recognized previously, the primary microrhombic fabric appears to reflect early mineral stabilization prior to significant burial compaction, although there is still disagreement on exactly how this occurred. Based upon an analysis of a Kharaib Limestone reservoir in Abu Dhabi, Oswald et al. (1995) argued for in situ dissolution and re-precipitation of originally metastable muddy sediments into stable microcrystalline calcite. As Moshier (1989a,b) had suggested earlier, they envisaged this occurring at slightly elevated temperatures (85o–105oF) and shallow depths (<200 ft) in a confined aquifer. In contrast, Al-Awar and Humphrey (2000) followed the hypothesis of Budd (1989) and suggested that the matrix microrhombic fabrics in the Shu’aiba reservoir of Ghaba North field in central Oman had developed in a confined to relatively open meteoric aquifer shortly after deposition.

  • Late Microdissolution. Further petrographic analyses confirmed that the reservoir pore system at Safah field has been critically enhanced by micro-intercrystalline dissolution porosity, as suggested by earlier work. During this period, similar late-stage micro-intercrystalline dissolution and permeability enhancement became more widely recognized throughout the region. Moshier (1989b) identified a late-dissolution event in the Shu’aiba reservoir of the Sajaa field and Wagner (1990) described selective leaching of microcrystalline limestones in Central Oman. Comparable late-stage dissolution was also interpreted by Al-Awar and Humphrey (2000) at Ghaba North and by Alsharhan et al. (2000) who recognized a late phase of secondary microporosity in Shu’aiba limestone reservoirs in the United Arab Emirates.

    As previously suggested, late-stage dissolution at Safah appears to have been controlled by microfractures and early matrix permeability distribution. The two episodes of fracturing recognized earlier, have been tentatively correlated with two well-established periods of regional structural readjustment (Figure 24). The first phase is thought to have occurred in the Turonian during the development of the Wasia-Aruma unconformity, when much of northern Oman and the United Arab Emirates was uplifted and partly exhumed. The second period has been attributed to the late Santonian-early Campanian event when the greater Lekhwair High was arched up as a peripheral bulge in response to allochthon emplacement (Boote et al., 1990). A significant part of the Wasia Formation was locally eroded from the Lekhwair High and North Lekhwair Arch. Following on from earlier analyses, microfractures formed at this time are thought to have acted as conduits for corrosive fluids expelled from the adjacent foredeep during the emplacement of the allochthon.

  • Intra-reservoir Faults. The influence of open faults and fractures upon reservoir compartmentalization and fluid flow became increasingly apparent during the later development of the Safah field. While low-displacement faults had occasionally been intersected by wells, the first direct evidence of open fracturing came from rapid breakthrough between widely spaced gas-injection and producing wells. While seemingly anticipated by earlier fault interpretations, these proved to be far subtler than previously suggested and were difficult to resolve seismically. Identification was initially dependent upon scattered well intersections and consequently was rather tentative. Inevitably, this led to significant disagreement about the frequency, distribution and origin of the fracturing.

Significant insights into the relationship between faulting, fracturing, and reservoir performance came from analyses of nearby mid- and Lower Cretaceous fields published at this time. Open-fracture systems were recorded in fields as far apart as Lekhwair on the platform (Arnott and van Wunnik, 1996) and the Natih (Whyte, 1995; Hitchings and Potters, 2000), Margham (Ernster et al., 1988) and Sajaa fields (Moshier et al., 1988) on the flanks of the Oman Mountains. They have had a very significant impact upon well performance, sweep efficiency and oil recovery. Fracturing intensity appears particularly severe in the high-relief structural traps flanking the mountains, but more subtle seismic scale (>10 m throws) and subseismic (< 5 m throws) faults and fractures were reported in the Lekhwair field (Arnott and van Wunnik, 1996). The orientation and distribution of these in-reservoir faults and fractures suggested a possible structural linkage with the better defined low-displacement fault array extending out from the Maradi Fault Zone and Natih-Fahud Horst (Figure 24). Based upon their regional setting and orientation, they have tentatively been interpreted as terminal splays of the Maradi fault system. This fault system had been attributed to Cretaceous transtension some time previously and, although primarily of Late Cretaceous age, there is more recent evidence to suggest at least local reactivation during intra-Miocene tectonic readjustment (Loosveld et al., 1996; Tschopp, 1967; Hanna and Nolan, 1989).

These nearby analogs provided a clearer perspective of in-reservoir faulting and fracturing at Safah. Following further seismic reprocessing, several additional prominent in-reservoir faults were identified in the field, characterized by small throws with pronounced northwest-southeast orientation very like the open faulting in the Lekhwair field (Arnott and van Wunnik, 1996). This similarity suggested that the subseismic-scale faults identified at Lekhwair might be more widely developed at Safah than had been anticipated and stimulated further analysis.

Current Safah Field Interpretation (2000)

The post-depositional history of Safah’s reservoir summarized in Figure 25 was developed from a synthesis of regional observations and interpretations combined with in-field data and analyses.

  • Although its relative importance remains uncertain, it is now widely accepted that the Shu’aiba platform at Safah experienced at least intermittent periods of subaerial exposure. While diagenetic overprinting has tended to obscure early meteoric fabrics, their repeated occurrence finally brought conviction despite their ambiguity. Extrapolation from regional observations suggests that meteoric dissolution was subdued because of the low relief of the platform, with aquifer recharge controlled by far-distant uplift rather than locally elevated topography.

  • Early meteoric leaching was followed by precipitation of non-ferroan calcite cements in a fresh-water phreatic environment, when much of the intergranular, intragranular and biomoldic pores were occluded.

  • Mechanical compaction and mineral stabilization of the mud matrix occurred shortly after deposition leading to the development of a microrhombic crystalline fabric. The similarity of the Shu’aiba δ13C and δ18O isotope values from Safah with those described elsewhere (Figure 11), suggested that its chalky texture must have formed in a similar way. Whether this occurred in an open meteoric diagenetic environment or in a closed system with near-marine formation fluids remains uncertain.

  • Further equant calcite cementation was followed with increasing burial, interrupted by an initial phase of microfracturing during regional late-Turonian uplift.

  • Following work by Prezbindowski et al. (1990), the enhanced micro-intercrystalline dissolution porosity and permeability is now generally attributed to a second period of microfracturing and introduction of corrosive fluids into the reservoir during emplacement of the allochthon wedge in the late Santonian to early Campanian. The causal link between primary facies and secondary dissolution, recognized in previous analyses, suggested that porosity-permeability distribution should mimic the stratigraphic architecture of the reservoir fairly closely. Consequently, with the development of a more comprehensive stratigraphic model of the field, it is expected that the organization, geometry, and continuity of reservoir flow units will be predicted with far more success than by earlier non-sequence-based reservoir models.

  • Early Tertiary hydrocarbon charge of the paleoclosure has been indirectly inferred from independent geochemical and structural analyses. While evidence is still ambiguous, later in-reservoir diagenesis at Safah is expected to reflect its subsequent mid-Tertiary structural tilting and charge history.

  • A final phase of reservoir enhancement has been attributed to a period of low-displacement faulting and fracturing during the Miocene. Seismic identification of larger in-reservoir faults has provided a critical template to infer the direction and orientation of subseismic-scale faults and fractures observed in well logs and cores. This has been used to develop a fault distribution model for the field. Its value in predicting fluid flow, sweep efficiencies and recoveries during the ongoing enhanced oil recovery program is still being assessed.

Hydrocarbon Charge

Regional Perspective

As development of the field continued, further geochemical analyses confirmed that the Safah crudes were indeed a mixture of oils from two quite different parent source rocks. However, the origin of the end-members remained controversial because of uncertainty about the relative importance of the δ13C isotopic data (Figure 16).

An exhaustive geochemical evaluation of oils in the area (Geomark, 1995) discounted the basal Silurian shale as a significant source at Safah. Instead, the crude in the reservoirs was interpreted as a mixture of high-maturity Infracambrian Huqf oils (such as those in Mezoon, Rawdah, and Wadi Latham) and a Mesozoic shale/marl-sourced oil similar to those in Upper Jurassic reservoirs of nearby Saudi Arabian fields. Concurrent geochemical studies had identified and described the following four significant and geochemically quite distinct Paleozoic oil families in Oman (Grantham et al., 1988; Guit et al., 1995; Al Ruwehy and Frewin, 1998; Terken, 1999; Terken and Frewin, 2000; and Terken et al., 2001):

  • the South Huqf Infracambrian oils in Dhofar,

  • Upper Infracambrian Q oils in south-central Oman,

  • North Huqf oils in the Fahud-Natih area of central Oman,

  • Silurian Safiq Formation-sourced oil (B type) at Sahmah field.

Contrary to the suggested Infracambrian contribution, a comparison of key biomarkers and isotopic character tentatively supported the earlier correlation between the isotopically lighter Safah crudes and a Silurian source (Figure 16). In part, this was based upon the absence of an homologous series of long-chain methyl-substituted alkanes (so-called X compounds) considered to be characteristic of Oman’s Infracambrian oils (Grantham et al., 1988; Terken and Frewin, 2000; Terken et al., 2001). In addition, the δ13C isotopic composition of the Safah oil lay well outside the range of both the South and North Huqf oil families (and of the Mezoon-Wadi Latham oils in particular) with only minimal overlap of their respective sterane distributions (Figure 16). In contrast, the isotopic and biomarker characteristics of the Sahmah crude (B type after Terken et al., 2001) matched the isotopically lighter Safah end-member relatively well.

With very few well penetrations, the distribution, quality and maturity of the basal-Silurian source rock (Qusaiba Hot Shale) in the greater Suneinah area could only be estimated indirectly from widely spaced control points flanking the southern part of the Arabian Basin (Figure 26). Peripheral wells, outcrops and distribution of Silurian-sourced hydrocarbons supported its presence across much of the Arabian Platform (Al-Husseini, 1991b; Aoudeh and Al Hajri, 1995; Stump and van der Eem, 1995; Bishop, 1995; Zumberge and Johansson, 1996; Droste, 1997; Milner, 1998; and Jones and Stump, 1999). However, it had evidently been eroded from a large part of Oman and the northern part of the United Arab Emirates during a period of Carboniferous to Early Permian doming and rifting (Blendinger et al., 1990; Mann and Hanna, 1990; Le Métour et al., 1995). Although absent from the Lekhwair High itself, regional projections suggested its erosional limit might lie just to the west (Figure 26) and certainly close enough to have sourced the isotopically lighter Safah crudes. Without direct control, the source-quality estimates are entirely speculative but large volumes of deep Silurian-sourced gas in the region hinted at its effectiveness. While obviously of very high maturity at the present time, the reconstructions of Bishop (1995) indicated that the area immediately offsetting the Lekhwair Arch was at peak oil expulsion during the Jurassic and Early Cretaceous. Consequently, it was argued that any Silurian-sourced oil at Safah may have re-migrated from a deeper paleoaccumulation nearby, perhaps in response to fault reactivation and uplift during the Miocene analogous to that observed elsewhere (Visser, 1991; Nederlof et al., 1995; and Terken and Frewin, 2000).

The origin of the isotopically heavier Safah crudes (Figure 16) also remained uncertain. During this period, Azzan and Taher (1995), Taher (1996, 1997), and Al-Suwaidi et al. (2000) had developed a more precise description of the Upper Shu’aiba source rock in the United Arab Emirates. Following redefinition of the Bab Member, it became clear that the organically rich facies first highlighted by Murris (1980) was confined to the basinal equivalents of the mid-Shu’aiba highstand sequence (mS2 and mS3, Figure 27). In contradiction of earlier reviews, Taher (1996, 1997) and Al-Suwaidi et al. (2000) argued that this was the primary source for the oil in the Shu’aiba and Kharaib reservoirs in central and eastern Abu Dhabi and, by implication, with the Safah and East Lekhwair oils. However, the isotopic overlap between the oils of the Thamama reservoir and the basinal Shu’aiba source-rich facies appeared limited (Figure 16). Furthermore, it was clear that at least some of these fields had been charged early during the later Cretaceous (Twombley and Scott, 1975; Oswald et al., 1995; Kirkham et al., 1996; Neilson et al., 1996) when the Shu’aiba generative area (Figure 27), as mapped by Azzan and Taher (1995) and Taher (1997), was still relatively immature. Ultimately, the mid-Shu’aiba source rock was considered to have been too far away and to have achieved maturity too late, to be a viable candidate for the oil at Safah.

Meanwhile, Hawas and Takezaki (1994) and others (Mohamed and Ayoub, 1992; Guit et al., 1995; Mohamed and Ennadi, 1995; Whittle and Alsharhan, 1996; Terken, 1999; and Terken and Frewin, 2000) continued to favor a Jurassic Diyab-Hanifa source for the Thamama reservoir accumulations in eastern Abu Dhabi and Lekhwair. Certainly, the geochemical similarity of these oils with deeper Jurassic reservoired crudes, at East Lekhwair (Grantham et al., 1998) and elsewhere, supported this correlation and suggested a Jurassic origin for the isotopically heavier Safah crudes (Figure 16). The distribution and maturity of Diyab-Hanifa intraplatform-basin source facies (Figure 12), described earlier by ADNOC (1984) and Loufti and El Bishlawy (1986), was broadly confirmed in later reviews (Thompson, 1995; Al-Suwaidi et al., 2000). Understanding of the basin’s stratigraphic architecture was greatly advanced by de Matos and Hulstrand (1995) and Al-Suwaidi et al. (2000), who demonstrated that the organically richest facies were developed in basin-center lowstand and distal highstand sequences of western Abu Dhabi. A review of their work suggested that thinner organic-rich intervals might extend farther east and interfinger with the distal part of the basin-margin carbonate cycles of eastern Abu Dhabi. Although now depleted (ADNOC, 1984), we would argue that such basin-margin source facies may have reached optimal maturity during the late Cretaceous and early Tertiary and would have been capable of sourcing the isotopically heavier oils at Safah, East Lekhwair, Mender, Qusahwira, and nearby in Saudi Arabia.

Current Safah Field Interpretation (2000)

More recent interpretations of the Safah field confirm that it is a stratigraphically trapped accumulation with a diffuse updip seal in the basin-margin facies of the Bab Member (Figure 28). Because of its poor reservoir quality and stratigraphic complexity, the oil-to-water transition is very extended and irregular. Earlier uncertainty about apparent differences in the elevation of the contact in different parts of the field was finally resolved when it became clear that these reflected subtle changes in the low-permeability fabric of the reservoir rock rather than in any significant changes in the height of the free-water level. The present-day oil-water contact (100% Sw) has been tentatively interpreted to be at a depth of 6,137 ft subsea with significant oil saturations (<50% Sw) generally terminating between -6,109 and -6,115 ft in the main part of the field and variably down to about -6,087 ft within the Upper Shu’aiba lowstand shoals. Residual oil shows occur in some areas and locally extend below -6,137 ft.

A paleostructural reconstruction of the top Shu’aiba in Eocene time (Figures 24 and 29) provides a useful insight into the field’s trapping geometry prior to mid-Miocene tilting. As suggested by the original exploration model developed 20 years previously (Figure 7), the accumulation appears to be located within the southern part of the North Lekhwair paleostructural closure, some way updip of its original crestal position north of Mkasa. The reconstructed closing paleocontour (Figure 30) coincides closely with the limit of the high GOR oils, and more approximately with the distribution of the residual oil shows in the northern part of the field. The distribution of the two parent oil families and their admixtures also appears to reflect the structural history of the trap. The δ13C isotope gradient observed in the reservoired crude (Figure 30) indicates that the heavier Diyab/Hanifa-like oils (subfamily A) are confined to the northern part of the field entirely within the paleoclosure. The isotopically lighter Silurian-like oils (subfamily D) are evidently restricted to the south, outside the paleoclosing contour and within the more recent stratigraphic closure formed as a result of Late Miocene structural tilting. Although these two parent oils have begun to co-mingle, this is clearly not yet complete and there is still a diffuse, transitional mixture (subfamilies B and C) between the two end members within the central part of the field.

While significant ambiguity still remains, this relationship between the evolving trap and distribution of contained hydrocarbons would appear to suggest the following (Figure 25):

  • Initial pre-Miocene charge of the larger paleoclosure by saturated high-GOR Hanifa/Diyab-sourced oils, with perhaps a quite extensive free gas cap to the north of the present accumulation.

  • Mid- to Late-Miocene tilting followed by re-organization and dispersal of the original gas cap.

  • Regional fault reactivation and breaching of a deeper Silurian-charged accumulation, followed by leakage, loss of dissolved gas, and vertical migration into the newly formed Shu’aiba stratigraphic closure above. Judging from the distribution of the partially mixed parent crudes, this late charge may have occurred contemporaneously with structural tilting before the Jurassic oils had re-equilibrated with the growing stratigraphic closure.

Analogous histories of multiple source and charging events with late in-reservoir mixing have recently been described by Nederlof et al. (1995), Al Ruwehy and Frewin (1998), Terken (1999), and Terken and Frewin (2000). As more information becomes available, late-Tertiary redistribution of deeper paleoaccumulations increasingly appears to be a common process throughout northeast Oman and presumably reflects subtle Miocene structural re-adjustment and uplift of the eastern foreland and adjacent Oman Mountains (Figure 24). In the Fahud field example described by Nederlof et al. (1995), the parent oils appear to be completely homogenized. It is not clear why mixing is still incomplete at Safah, but it is possible that more rapid in-reservoir cross-flow may have been slowed by the low matrix permeabilities and oblique stratigraphic barriers and baffles that are characteristic of the field.

A phase of late gas charging and possible flushing has also been recognized in the greater Suneinah area and has been proposed as an alternative explanation for the gas caps at Safah (1995). Certainly, there is good evidence of gas flushing in some of the smaller accumulations to the east and downdip of Safah. This is thought to reflect Miocene uplift-controlled gas exsolution and fault leakage from deep, thermally cracked source rocks or paleoaccumulations. Whether this occurred at Safah is still uncertain, although the distribution of the gas caps (Figure 30) and their association with high GOR oils tends to favor an earlier Jurassic origin from the west.

Reserve Estimates

During the earlier part of the development period, hydrocarbon-in-place assessments rose modestly to approximately 820 MMbbls oil and 730 BCF gas. However, the gas injection program initiated in 1994 proved very successful and recovery estimates increased significantly by 21 to 22 percent during this time (Figure 4).

Oil-in-place estimates were increased in 1997 to 880 MMbbl (with a high side of 1,125 MMbbl) by acknowledging the low oil saturations within the transition zone (>50% Sw). This larger estimate was remarkably close to the upside appraisal assessment calculated in 1985 (Figure 4). A successful trial waterflood program started in 1999 offered the possibility of significantly higher water saturation cut-offs than previously assumed. As a result, reserves (recoverable oil) are now estimated at between 380 and 430 MMbbl with the total oil-in-place assessed at 1,077 MMbls on the basis of a recovery factor of 35 to 39 percent.

PREDICTING FROM INCOMPLETE INFORMATION—CONCLUSIONS

Dott (1998) argued that geological reasoning must rely upon qualitative, circumstantial evidence rather than direct experimentation, because of its time-bound historical reality. Of necessity, explanations have to be deduced from past events by synthesizing observations into a narrative logic most consistent with the available data. Such narratives provide a context or perspective in which the observations must make coherent sense. However, these can only represent just part of a spectrum of possibilities. The precision with which they describe a geological reality is limited by the quantity and quality of the data available, while the accuracy with which they describe it is constrained by the way the data was selected and by the conceptual framework in which it is evaluated. Data collection and sorting is itself not merely a simple objective process. Geological information is often conflicting and ambiguous especially in complex areas, with gaps to leap and irrelevant data to ignore. Data selection and evaluation is often biased by the preconceptions of the observer. Any subsequent synthesis requires a conceptual framework with which to judge what explanations might work and what will not (Miall and Miall, 2001). Such perspectives are themselves historical, reflecting the conceptual paradigm fashionable at the time and are shaded by them. Data selected and synthesized under one set of preconditions may point to a quite different interpretation than it might under another. At Safah, such preconceptions had an enormous impact on the decision to test the original play concept, in the subsequent appraisal-drilling program and in defining field-management strategies during its later development.

During its history, several historical narratives were developed to describe Safah, each attempting to best define its geological reality with the available data. The exploration concept that led to the successful Safah-1x discovery well was novel for the area. Certainly, it was only partially correct. The updip trapping mechanism was far more complex than anticipated and its reservoir quality far poorer than suggested by the favored analog. The possibility of dual charging was not even considered. Clearly, predictive accuracy at this level was tempered by limited information. However, those more general predictions derived from a broader more global perspective were very successful. Interpretations of the paleostructural history, migration focusing, charge timing and the effect of late tilting, all proved prescient.

The appraisal program resolved the most critical pre-discovery uncertainties. Trap, reservoir, and charge were all confirmed, but as increasing amounts of information was acquired the evaluation became more exacting and required ever more focused predictions. Ambiguities about the reservoir architecture arose out of seeming conflicts between independent geological observations. Its subtle low-relief stratigraphy with few distinctive marker horizons defied easy correlation. Rock isotope data apparently precluded Upper Shu’aiba subaerial exposure and, in combination with contra-stratigraphic PVT property and oil isotope gradients, frustrated early consensus on an appropriate static reservoir model.

Without the support of an appropriate regional perspective, the available information was synthesized in different ways, reflecting quite different preconceptions and biases. It was only when new regional syntheses provided an improved conceptual framework that these contradictions were resolved. The credibility of the revised mid-Shu’aiba highstand/Upper Shu’aiba lowstand stratigraphic model was supported by its apparent ubiquity throughout the region. The importance of the Safah rock-isotope data was discounted by comparison with similar sequences elsewhere. The possibility of dual sourcing came from a full appreciation of the regional hydrocarbon environment and its recognition proved critical in establishing the integrated two-phase entrapment model which finally resolved the apparent conflict between stratigraphy and fluid-property data. More advanced technology had a profound influence on understanding the field by expanding into areas of enquiry not otherwise accessible. Improved seismic, more sophisticated geochemistry and petrographic techniques all made it possible to ask new questions and rephrase old ones. However, the issues they raised were still geological and could only be absorbed and utilized within the context of an historical narrative.

Based upon our experience at Safah, we would argue that the accuracy of any geological prediction very much depends upon the conceptual framework in which it is evaluated and tested. While direct comparisons with nearby analogs tend to be limited by their emphasis on the typical or established case, they are of tremendous value in deciding what works and what does not when broken down into their component parts. Although the Safah field so far appears unique in the larger South Arabian Basin, the stratigraphic, structural, petrographic and geochemical concepts used to describe it were all borrowed and tested against geological explanations used elsewhere in the region to account for similar geological phenomenon.

As the Safah field development program matures, this history of constantly evolving geological perceptions is sure to continue. New data and concepts will continue to supplant old ones, while old ones might regain favor in the light of new information. We believe that successful geological predictions will continue to come from new patterns and linkages of data interpreted and tested with a broad narrative construct. It is only by assessing such predictions against a synoptic or global perspective that it is possible to judge their relative merits with any confidence.

ACKNOWLEDGMENTS

The Ministry of Petroleum and Minerals, Sultanate of Oman, and the management of Occidental and Exploration Company, Occidental of Oman, and Fortum (E&P) B.V. are thanked for their permission to publish this article.

Many imaginative geoscientists have participated in the Safah story since 1980, suggesting, defending and not infrequently opposing the many and varied ideas and concepts proposed to explain its ambiguous and perplexing geology. We are fortunate to have had the opportunity of sharing in their spontaneity, enthusiasm, and friendship over the years, and gratefully acknowledge their intellectually courageous and always creative attempts at understanding the field:

A.S. Ahmet, P.M. Austin, B. Barrick, W.C. Benmore, C.J. Blom, D.M. Bliefnick, F. Bourgeois, R.C. Buruss, J.D. Burgess, A.J. Burrows, S. Burns, J.A. Carver, V. Childs, N. Clayton, M. Cleveland, D.R. Clowser, D.W. Cripps, B. Dockweiler, J. Elliot, M.H. Fischer, C.G. Fisher, S.H. Frost, C.L. Funk, M.H. Girgis, G. Graham, S. Gustav, S.D. Harker, T. Harland, P.M. Harris, M. Hawas, A.J. Holt, N. Hulse, N. Hummer, D.G. Kelland, G.W. Kendall, D.C. Kitson, H. Lee, D.R. Martin, P. Martinez, C. Mather, P. Mathieu, C. McDonald, G. McGann, J. Meyer, C.A. Miles, D. Moore, C. Noseda, P. Nygreen, C.R. Oban, R.K. Olson, S. Palmer, C.A. Parker, T. Perkins, J.R. Phillips, D. Prezbindowski, R. Pulley, G. Rees, M. Raven, D. Renoult, B. Roosjen, V. Russell, A. Saller, M.J. Sauer, G. Scott, G.A. Seiglie, A. Seliem, H. Serim, B.L. Shaffer, J. Sherwood, H. Silsmets, G.M. Smith, R. Smith, L. Smith, M.P. Stark, S. Stearns, C. D. Todd, R.H. Vaughan, F. Vlierboom, R. Waite, E.F. Watson, M. Williams, P.L.H. Worley, S. Wykes, D. Yap, and J. Zumberge.

We also very much benefited from the active and constructive comments of Nigel Banks, Bob Dockweiler, David Kelland, Rob Kirk, Conrad Maher, Gerry McGann, Bob Olson, Dennis Prezbindowski, Peter Sharland, and John Smewing on earlier versions of the text. We thank Diana Monckton for her patient editorial assistance while preparing the manuscript, and Liliana Indaburu and David Lombard for creating the accompanying illustrations. The text also benefited from many pertinent suggestions by Henk Droste and an anonymous reviewer. Text editing and the design and drafting of the final graphics was by Gulf PetroLink.

ABOUT THE AUTHORS

David Boote is a Consultant Geologist. He had 25 years experience in international new venture exploration with Occidental Oil and Gas Company in various technical and supervisory positions, responsible for basin evaluations, play analysis, prospect generation and appraisals in many parts of the world, including Oman, Yemen, Australia, the Philippines, Algeria, Libya, UK, and California. Since leaving Occidental in 2000, David has been active as an independent consultant in regional stratigraphic and hydrocarbon petroleum system syntheses.

davidboote@elsyngeroad.fsnet.co.uk

Duenchien Mou is a Geological Consultant. He has 24 year of international new venture exploration in various positions, including Worldwide Exploration Chief Geologist and overseas Exploration Manager in the Netherlands and Colombia with Occidental Oil and Gas. He was involved in the geological evaluations of a wide variety of sedimentary basins and petroleum provinces worldwide with emphasis on the Middle East, Far East, North Sea, and South America. Duenchien is now managing Exploration Consulting International (ECI) to provide independent technical advice on international new ventures with a focus upon petroleum system analysis, prospect generation, and risk assessment.

dmou@bak.rr.com