Tilted original oil/water contact in the Arab-D reservoir, Ghawar field, Saudi Arabia

In fully buried reservoirs, trapped fluids are generally in a static condition prior to production, and the Original Oil/Water Contact (OOWC) would normally be horizontal. However, naturally occurring tilted OOWCs have been described. The most commonly accepted explanations are ‘frozen-in’ diagenetic trapping combined with late-tectonic tilting (‘forced’ static tilt), or regional hydrodynamic aquifers (dynamic tilt).

Wilson (1977) introduced the idea that diagenetic porosity reduction in the aquifer combined with tectonic tilting may create tilted OOWCs. Yeats (1983) and Carlos and Mantilla (2000) interpreted tilted OOWCs as a result of tectonically induced rapid development of structural folding or tilting in reservoirs of low absolute permeability. Willingham and Howald (1965), Pelissier et al. (1980), Wells (1987, 1988), Winterhalder and Hann (1991), Beckner et al. (1996), Gauchet and Corre (1996) and Luebking et al. (2001) relied on aquifer hydrodynamics to account for tilted original contacts.

For a geological formation that crops out at a high topographic elevation, rainwater infiltrates into the aquifer. If the aquifer system is also outcropping at a lower elevation, differences in the hydraulic head will induce water flow. Where hydrocarbons are trapped deeper in the sedimentary basin, the flow in the aquifer leg may lead to the dynamic tilting of the OOWC.

Dickey (1963), Dickey and Soto (1974) linked aquifer activity with chemical composition at the scale of the sedimentary basin, and showed that highly saline brines are characteristic of static aquifers. Of special importance for this paper was the concept introduced by Bond (1973, 1975) of aquifers in static equilibrium even though changes in hydraulic head were measured. The apparent paradox was explained by variable salinity at the regional scale. To the best of our knowledge, this concept had not been applied to tilted OOWC prior to the work of Stenger (1999).

We propose to interpret the tilted OOWC in the Ghawar Arab-D reservoir by the combined effects of changes in oil and water densities in and around the Ghawar field. After discussing the regional setting, the tilted OOWC in Ghawar Arab-D reservoir will be described through field observations. A discussion on the static or dynamic nature of the tilted OOWC will review different mechanisms and interpretations. Finally, conclusions on the origin of the tilted OOWC will be submitted together with a brief review of implications for the on-going development of the southernmost area of Ghawar. As indicated by Aramco (1959) and discussed by Stenger (1999) and Stenger et al. (2001), the tilted OOWC in Ghawar does not lend itself to a straightforward classification.

The Arab-D carbonate reservoir of the Upper Jurassic Arab Formation in the onshore Ghawar field was discovered in 1948. Following further separate discoveries along the structure’s main axis, five production areas were quickly identified as parts of the giant Ghawar oil field (Figure 1): from north to south they are ‘Ain Dar, Shedgum, Uthmaniyah, Hawiyah and Haradh. At the Arab-D level, the field is a NNE-trending composite anticline 230 km long and about 30 km wide (Figure 2a). The gently dipping crestal region is composed of several sub-parallel axes. The anticline is asymmetric and fairly steep-sided (up to 10° dip). In the southernmost extension of Ghawar (southern part of South Haradh) the west flank is steeper (Figure 3). Farther north, for example in central Haradh and Uthmaniyah, the eastern flank is steeper. Figure 2b is a regional depth map of the top Jurassic.

According to Wender et al. (1998), the growth history of the Ghawar structure consisted of the following four main phases: Hercynian Orogeny (Carboniferous); Zagros rifting (Early Triassic); Early Alpine Orogeny (Late Cretaceous); and Late Alpine Orogeny (Tertiary). Post-Jurassic tectonic activity was generally mild and limited to the multistage rejuvenation (uplift and erosion) of the Ghawar structure—a part of the greater En Nala Anticline—that is bounded by major N-trending Hercynian basement faults.

The ‘Ain Dar and Shedgum areas went on stream in 1951 and development progressed southward by stages. Due to the lack of aquifer support, peripheral gravity water-injection was started in the late 1960s to maintain the reservoir pressure at an adequate level. Since the early 1980s, powered seawater injection has replaced gravity injection in a bid to conserve freshwater resources.

During the last fifty years, thousands of vertical wells have been drilled on 1-km spacings. Most of these wells have open-hole completions. In the last five years, the use of horizontal drilling has intensified, especially in the southernmost area where the degraded reservoir quality limits the productivity of vertical wells. At the time of writing (2002), the Arab-D reservoir development is reaching the southernmost Haradh area (Figure 2a). North Haradh has been producing since mid 1996 at the established plateau rate, and the Central Haradh drilling is scheduled for completion in early 2003.

To support the on-going field development and management, Saudi Aramco is applying the latest technologies in reservoir characterization, modeling, and 3-D visualization. For instance, a modeling challenge lies in the sheer size of the Ghawar Arab-D reservoir that requires several reservoir sector models to be maintained. Thanks to the development by Saudi Aramco of POWERS—a reservoir simulation software based on massive parallel processing—a single multimillion-cell model may now be used to describe the fluid-flow mechanisms of entire giant fields (Dogru et al., 2001). Meyer et al. (2000) and Cantrell et al. (2001) proposed innovative interpretations for the origin and distribution of dolomitic super-permeability (super-k) intervals in the Ghawar Arab-D reservoir that are responsible for anomalously high production rates. According to Cantrell et al. (2001), their distribution is not correlatable to the rock facies but rather to a degree of structural control. Al-Shahri et al. (1998), Al-Ajmi et al. (2001), and Phelps et al. (2000, 2001) have evaluated optimal production strategies for the Arab-D reservoir by taking into account such geological complexities as super-k intervals and fracture clusters/swarms.

In Ghawar, the Arab-D reservoir is divided from top to bottom into lithostratigraphic Zones 1 to 4, with Zones 2 and 3 being subdivided into subzones A and B (Figure 4). The best reservoir quality is in Zone 2, described as having been formed in a shallow-marine, high-energy environment. The sharp decrease of the Arab-D reservoir properties below the middle of Zone 3A, has had an adverse effect on the oil saturation as recorded by logging tools (Figure 5). As porosity and permeability drop, open-hole readings of water saturation increase rapidly to 100 percent due to large capillary pressures (pore-size effect). In the Ghawar field, the pay porosity cutoff is 4 percent.

From north to south, the quality and thickness of the Arab-D reservoir steadily decrease and the productivity of vertical open-hole producers drops four-fold (Figure 6). At the time of writing (2002), all known static and dynamic data on Ghawar indicate lateral reservoir continuity across the entire field. In particular, fault throws mapped from 3-D seismic are not large enough to create reservoir compartmentalization. However, the Arab-D reservoir is a seismic reflector of variable quality and lineaments derived from the 3-D seismic are generally better defined at shallower or deeper levels. Image logs have shown that natural fracturing is present throughout the Arab-D reservoir in the Ghawar field (e.g. Phelps and Strauss, 2000). Borehole breakouts and fractures have been analyzed in vertical and horizontal wells. The azimuth of the (present-day) maximum horizontal in situ stress varies from N60°E in Uthmaniyah to N110°E in the south (Figure 7a). In Haradh, natural fracturing shows a complex tectonic history with the existence of three main families of fractures having trend directions of N10°E, N80°E, and N130°E (Figure 7b).

In eastern Saudi Arabia, most of the units above the Arab Formation are composed of carbonate rocks. Thus, by knowing the thermal conductivity of the upper units, it is possible to compute an average geothermal gradient from the near surface to the Arab Formation, as the thermal conductivity will be fairly uniform. Although no direct heat flow data are available, the average geothermal gradient can be calculated from the difference between the reservoir temperature and a depth below the Earth’s surface where the change in temperature due to solar radiation is negligible. It is assumed that the baseline temperature is 50^oF at a depth of 30 ft. A total of 110 geothermal gradients was used to construct the geothermal gradient map (Figure 8a). The map shows a high of 3.2°F/100 ft on the Dammam Dome and a decreasing geothermal gradient from 2.5°F/100 ft in northern Ghawar to 1.8°F/100 ft in the south.

From 1970 to 1983, the Haradh area was producing by primary depletion. This caused the reservoir pressure and temperature on the western flank of Central Haradh to fall by 700 psi and 34°F, respectively. From 1983 to 1990, the Haradh sector was mothballed following a sustained drop in the international demand for oil and the reservoir temperature recovered to its initial value. A 1991 temperature map of the Haradh area shows a remnant area of lower temperature on the west flank (Figure 8b).

The salinity of the regional Arab-D aquifer is a function of the Arab-D burial depth and geothermal gradient as shown by the good coefficient of correlation (r² = 0.86) using a two-parameter linear regression (Figure 9). However, a poor match exists between the average measured (42,000 ppm Total Dissolved Solids—TDS) and calculated (102,000 ppm TDS) salinity on the west flank of Central Haradh. The regression also overestimates the measured salinity on the western flank of southern Hawiyah. A closer look at the Ghawar aquifer salinity shows a strong contrast between the isolated western and eastern aquifer legs (Figure 10), although the scarcity of early salinity data in and around Ghawar does not allow a definitive picture to emerge. The largest salinity contrast is across Central Haradh. This is an increase from 30,000 ppm TDS in the west to 152,000 ppm TDS in the east. In 1981, a repeat formation test run in a southernmost Ghawar delineation well, indicated a water-pressure gradient of 0.457 psi/ft, equivalent to a salinity of 75,000 ppm TDS. On average, the aquifer water-pressure gradient decreases from 0.50 to 0.45 psi/ft from north to south Ghawar.

In a bid to better understand the salinity change in Ghawar, stable isotopes of oxygen (δ¹⁸O) and hydrogen (deuterium, δD) have been quantified for several aquifer samples. W.J. Carrigan and S.H. Al-Sharidi (unpublished Saudi Aramco memo HSD 188-98, 1998) showed that Arab-D aquifer water samples from the western flank of Central Haradh fall on the Pleistocene meteoric water line (6,000 to 20,000 years ago) and are clearly different from the δ¹⁸O/δD isotopic range of the Arab-D Formation water (Figure 11). In Central Haradh, a water sample was taken recently from the shallower Wasia Formation (Middle Cretaceous) in a water well. The isotope analysis of this sample is remarkably close to the anomalous Arab-D water sample taken from the west flank of Central Haradh (Figure 11).

From north to south Ghawar, the in situ oil pressure gradient increases from 0.302 to 0.328 psi/ft (Figure 12) in relation to a total gas/oil ratio decrease from 636 to 346 standard cubic feet/stock-tank barrel. A variable concentration in methane is primarily responsible for the observed change in the oil pressure gradient. Using a simple linear regression, we noted a good correlation coefficient (r² = 0.79) between the regional geothermal gradient and the changes in oil density throughout Ghawar (Figure 13). In our interpretation, the geothermal gradient is the main control on the oil density distribution in Ghawar.

The initial Ghawar Arab-D reservoir pressure of about 3,215 psi at a datum depth of 6,100 ft, was close to the hydrostatic range. In ‘Ain Dar, the original reservoir pressure in 1948, in discovery well ANDR-1, was 3,204 psi at a depth of 6,100 ft. However, this value was reportedly influenced by production in the nearby Abqaiq field and was less than the original estimate of 3,226 psi at 6,100 ft (Aramco, 1959). Discovery well HRDH-1 was production tested in April 1949, with a pressure build-up of 60 hours. The initial reservoir pressure was 2,925 psi at −5,200 ft. After correcting for the reservoir oil gradient of 0.323 psi/ft, this gave the initial reservoir pressure in Haradh as 3,216 psi at 6,100 ft. A map of the Haradh pre-production pressure measurements shows a weak east-west trend (Figure 14).

Due to capillary effects, the Oil/Water Contact (OWC) is distinguished from the Free Water Level (FWL). Whereas the former may be considered to be any depth where oil is mobile in the transition zone, the latter is synonymous with the zero-capillary pressure surface (Figure 15). In good reservoir sections where capillary pressures may be neglected, OWC and FWL are superimposed. The purpose of this paper is to investigate the Ghawar Arab-D FWL as observed in clean reservoir sections. However, for the sake of simplicity, we will use the OWC/OOWC acronyms in the following discussion.

Early in the development of the Ghawar field, Aramco became suspicious that the OOWC picks on open-hole logs (Figure 16) were surprisingly close to the top of the low-permeability reservoir section (Figure 17) (unpublished Aramco report, 25-N, File Gh-1.2, June 1953). For the sake of visual clarity, the OOWC surface is interpolated over the whole Ghawar field area although the aquifer physically exists only along the edge of the reservoir. This raised legitimate doubts as to the validity of the contact identification. Further complications were added to the definition of the OOWC, identified in early studies as, “the lowest level above which clean oil would be produced from a well during a drill stem or production test”. From a comparison of Drill Stem Test (DST) data with induction/electric well log analyses, Aramco found that this “producing OOWC coincides, in good quality Arab-D rock, with an interpreted open-hole log water saturation of 35%”; that is, close to the top of the transition zone. DSTs in the good reservoir sections produced only clean oil whereas those in the tighter sections produced salty crude. It was also noted that intervals tested as producing salty crude during DSTs produced clean oil for several years after completion. The salt content measured during the DSTs had been interpreted as interstitial formation water—not aquifer water—being mobile due to the larger drawdown imposed on the Arab Formation. In those days, the salt content of the crude oil was a major cause of disruption to the catalytic process used in the refineries.

Due to operational constraints, a precise delineation of the OOWC was not always a primary consideration as, in general, the interface was expected to be horizontal—and to recognize it once was enough, providing that the reservoir was not compartmentalized. In Ghawar, several factors have affected the delineation of the tilted OOWC in the Arab-D reservoir. For example, the OOWC is only present at the edge of the oil accumulation. This, combined with the steep flanks (for instance in the west in Central Haradh) (Figure 3), has made it difficult to drill delineation wells that intersect the OOWC in areas where it is unaffected by large capillary pressure effects, such as in Zone 2.

In the early days of the field appraisal, the increase in water saturation in Zone 3 was considered to be due to the proximity of the water leg, regardless of the quality of the reservoir rock, and this led to wrong picks for the OOWC. For instance, in the absence of any other information, the OOWC might be picked at a depth of 6,470 ft in well ANDR-A (west flank) and at 6,580 ft in well SDGM-B (east flank), which would result in a 110 ft west-to-east tilted OOWC (Figure 18). A careful review of the open-hole logs in ‘Ain Dar and Shedgum (North Ghawar) showed that a west-east tilt of the OOWC was almost impossible to determine. Another difficulty that arises when trying to assess the OOWC tilt comes from picking an Oil-Down-To (ODT) in wells already affected by water encroachment caused by updip production. SDGM-C was one of the key wells to demonstrate the existence of a west-east component of the tilt in North Ghawar. However, the well was drilled in May 1969, or 11 years after production started, and was completed in an area of extensive updip production. Looking now at the open-hole logs, some lagging oil can be spotted below the picked ODT. Therefore, if the OOWC is taken to be below the deepest lagging oil zone, the west-east tilted fluid contact cannot be substantiated.

DST and production data, core descriptions, and open-hole logs were integrated to reduce the uncertainty attached to each type of data. In this study, it was proposed to look for the FWL depth in the early Ghawar flank wells. Where the identification of the FWL was obscured by the presence of low-permeability rocks, the ODT and Water-Up-To (WUT) depths were picked. The OOWC review started with the examination of the open-hole logs from all possible flank wells irrespective of their drilling dates. This gave 102 wells on the west flank and 93 wells on the east flank (Figures 19a,b). As expected, some scattering was apparent when plotting the ODT and WUT picks against the depth of the good reservoir section (4% porosity cutoff). Although several ODT and WUT picks were superimposed, or close to the base of the good reservoir section, they showed with a reasonable confidence that the OOWC shallows from a depth of 6,650 ft in the north (‘Ain Dar) to 6,400 ft in the south (Haradh). In Central Haradh, the west flank OOWC was shallowest at 5,800 ft whereas, at the same latitude, the east flank OOWC was located between 6,532 ft and 6,631 ft. This gave a OOWC downward tilt from west to east of between 700 and 800 ft.

In order to ascertain further the trends identified above, a subset of flank wells was chosen with the proviso that the well had been drilled before any significant cumulative production had occurred in the area. The subset consisted of 19 wells on the west flank and 15 wells on the east. In addition to open-hole logs, 12 of the wells had DST’s, two had Repeat Formation Tester logs, and 18 were cored. The subset confirmed the OOWC trends noted earlier with the shallowest OWC in well HRDH-A (Figure 20a) and the deepest in well ANDR-D (Figure 20b).

This extensive revision confirmed earlier studies dating back to Aramco’s early field appraisal (Aramco, 1959). The elimination of uncertain evidence allowed the tilt geometry to be clarified. The tilt of the Ghawar Arab-D OOWC has the following major components:

• a field-wide SSW-NNE low-gradient tilt with a 200 ft deepening of the OOWC from south to north Ghawar, a distance of 230 km; and

• a localized W-E high-gradient tilt with up to 800 ft deepening of the OOWC from west to east Central Haradh, a distance of 25 km.

In the past 20 years, several concepts have been proposed to explain tilted OOWC worldwide. We reviewed several of these that may explain the tilted OOWC in the Ghawar Arab-D reservoir. In order of decreasing importance for our discussion they are as follows:

• fluid properties and static equilibrium;

• dynamic aquifer;

• tectonic tilting and frozen-in diagenetic traps; and

• Earth’s gravity and inertial forces.

Following the geothermal gradient, the reservoir oil and water densities steadily change across the Ghawar field. As the reservoir temperature decreases from north to south, the oil density increases by 6 percent and the water density decreases by 10 percent. The proposed explanation is that at higher temperatures, water salinity (and thus water density) is higher. Conversely, reservoir oil density is lower at higher temperature due to thermodynamic segregation. The existence of opposite trends in water and oil densities implies that the weight of the oil and water column changes throughout the field. In the case of a horizontal OOWC with constant fluid density, the isobaric surfaces related to the pressure exerted by the oil column are horizontal. Given the variable oil and water densities as measured in Ghawar, these surfaces are tilted. Stenger (1999) suggested the following:

• At some distance from the Ghawar field, isobaric surfaces must return to the horizontal in the Arab-D Formation by mirroring the vertical segregation of water salinity at the basin scale. Chiarelli (1973) provided evidence for this by describing the Jurassic aquifer of Saudi Arabia as being a rejecting and passive system in contrast to the recharging and dynamic aquifers of Cretaceous and Eocene age. Therefore, and as shown by the available data (Figure 9), aquifer water density (salinity) should be mainly a function of reservoir depth and temperature in the Arab-D reservoir.

• Aquifer waters of different salinities or temperatures do not easily mix in a porous medium. This leads to the possibility of fresh water introduced from shallower aquifers as ‘floating’ on top of the more saline water.

Dickey (1963) formulated similar ideas when discussing the general topic of underground waters and oil exploration. He later repeated the idea that a tilted OOWC does not necessarily imply hydrodynamics, especially in basins where the aquifer salinity shows high contents of chlorides, calcium, and total dissolved solids that indicated stagnant aquifers (Dickey, 1968, 1988). Bond (1973) defined a variable density aquifer as one in which the density of the interstitial water varies from point to point. He concluded that observed (hydraulic) gradients of a few feet per mile in the saline part of the aquifer probably had little or no significance with respect to flow. Aramco (1959) indicated that, “there may be a possibility of large-scale vertical salinity stratification” in the Arab-D aquifer.

It is possible to calculate the equilibrium of the oil and water columns across the structure based on the above assumptions (Stenger, 1999). At the northern end of Ghawar (salinity 225,000 ppm TDS), the OOWC is 6,650 ft deep, whereas the calculated OOWC (from an average salinity of 65,000 ppm TDS) is 6,368 ft at the southern end of the Haradh sector. This is within the OOWC approximation of 6,420 ft, as observed in the southernmost Haradh delineation well (HRDH-J) by RFT (Figure 21). It is important to note that if a salinity of 35,000 ppm TDS is used for Haradh, as observed on the west flank of Central Haradh, the shallowest OOWC would be at a depth of 5,835 ft. This is within the observed estimates of from 5,863 ft (HRDH-A) to 5,820 ft (HRDH-B) for the west flank of Central Haradh. Calculated and observed OOWC tilts show similar trends (Figure 22), although on a well-by-well basis some discrepancies are irreconcilable due to local and minor irregularities not accounted for in this interpretation. The map of the calculated OOWC (FWL) is from a simple extrapolation of the equilibrium calculation performed on the N-S cross-section. A similar exercise in the west-to-east direction would require more data on the oil and aquifer water densities than is presently available. It is important to note that any discernable change in water salinity between the aquifer legs will lead to a tilted interface. Appendix A gives an example of an analytical calculation for a theoretical structure.

Hydrodynamics affect water-bearing reservoirs outcropping on structural highs where rainfall controls the recharging process. When sufficient reservoir permeability combines with a discharge area (springs), fluid flow can be sustained from the structurally high recharge area to the structurally low discharge area. Water flowing in such aquifers usually has low salinity and is young. According to Dickey and Soto (1974), in hydrodynamic aquifers “meteoric waters may be defined as waters that have been part of the hydrologic cycle recently, geologically speaking”. In recognized hydrodynamic aquifers, the existence of fluid pressure isopleths is consistent with the direction of aquifer flow and the tilt of the contact. Hydrocarbon pools with bottom dynamic aquifers usually have a high degree of natural pressure support throughout their production history, as discussed for instance by Winterhalder and Hann (1991).

In the case of the Arab-D reservoir of Ghawar, V.V. Valleroy (unpublished Exxon Production Research Letter ER-82-25, 1982) proposed that hydrodynamics caused the tilted OOWC (Figure 23) and that hydraulic pressure gradients from 1.8 to 10 psi/km were necessary to explain the observed tilts. A regional Arab-D reservoir simulation model reproduced the tilted OOWC by an aquifer flow of 5,000 barrels of water per day over a period of 20,000 years. The results showed that the average pressure gradient was about 8 psi/km between ‘Ain Dar (North Ghawar) and Haradh (South Ghawar). However, this translated into an original pressure difference of 1,300 psi before production started (Figure 24a) whereas pre-production pressures measured in exploratory wells in ‘Ain Dar and Haradh did not show any pressure gradient that substantiated a dynamic aquifer flow from south to north. Another simulation run attempted to reduce the pressure gradient by lowering the injection rate (100 barrels of water per day over a period of 135,000 years). Although the pressure difference between ‘Ain Dar and Haradh was reduced to 200 psi, the magnitude of the tilted contact could not be reproduced (Figure 24b). In Central Haradh, a minimum pressure difference of 90 psi would be necessary to maintain the observed W-E tilted OWC. Again, the initial reservoir pressures measured in Haradh did not show such a difference.

Clearly, slow fluid movement occurs naturally in a reservoir over geological time. However, it is of a very different magnitude to the situation in a dynamic aquifer where significant flow and pressure gradients are obvious on a production time scale. In retrospect, the lack of aquifer support in the Arab-D reservoir should have been another indication that hydrodynamics was not a robust explanation for the tilted OOWC in Ghawar. In addition, the issue of the discharge area located northeast of Ghawar on Figure 23 and apparently draining into the central low of the sedimentary basin, was not solved.

Tilting or folding of subsurface structures has been proposed to explain the tilted OOWC in the oil fields of southern California (Yeats, 1983). Low matrix permeability normally requires extensive periods of time to recreate the gravitational fluid segregation, but the tectonic activity in California increases the likelihood of such occurrences. Carlos and Mantilla (2000) claimed to have evidence of a tilted OOWC caused by tectonics but their case was weakened by a lack of pre-production pressure data, a strong bottom aquifer drive, and reservoir permeability of about 300 mD.

The Ghawar area shows little evidence of sub-recent tectonic activity. In Haradh, Aramco (1959) stated that, ”the only suggestions of disturbance, and these are virtually unsupported, are found in the Wadi Sahba area (see Figures 27 and 30) where drainage, originally toward the north and northeast may have been diverted toward the east and perhaps southeast. The timing of this tilting should be early Pleistocene and perhaps even late Pliocene. Regional slopes are so gentle, however, that effects on the oil field must have been very slight.”

For the following reasons we agree with Aramco (1959) that tectonic tilting cannot explain the observed west-to-east tilt of the OOWC in the Haradh area.

• The shallower OOWC observed on the west flank of the Haradh sector is extremely localized and the contact steepens rapidly in all directions (Figure 25).

• The relatively high reservoir permeability in the Arab-D of Haradh (300 mD), cannot equate with a long re-equilibration period—with structural tilting dated as Pleistocene, equilibrium should have been achieved within a few thousand years and the OOWC should now be horizontal.

• T.R. Pham and A.S. Al-Muhaish (unpublished Saudi Aramco ‘Udhailiyah Reservoir Management Division Internal Memo, 1998) interpreted residual oil as being present in dolomitic stringers below the current OOWC on central Haradh’s western flank. This would mean that oil was previously present at a deeper level than is observed today on the west flank and would contradict the possibility of clockwise tectonic tilting to explain the observed tilted OOWC in Haradh.

• The possibility that the OOWC was ‘frozen-in’ following the diagenetic destruction of porosity in the aquifer does not seem likely for Ghawar.

So far, no drastic porosity or permeability reductions have been recorded in the Arab-D aquifer, and the reservoir pressure has been maintained efficiently through a peripheral water injection scheme since the late 1960s. Furthermore, the Abqaiq and Harmaliyah fields located north and east of Ghawar (Figure 1) are in hydraulic communication through the Arab-D aquifer leg.

The acceleration due to gravity (g) is not uniform over the Earth’s surface (Figure 26) and is used in the traditional equation for determining hydrostatic head. The main axis of the Ghawar anticline is perpendicular to the isogals and has a minimum acceleration of 32.113 ft/s⁻² (9.788 m/s⁻²) at Haradh in the south and a maximum close to 32.119 ft/s⁻² (9.790 m/s⁻²) at ‘Ain Dar in the north. The acceleration due to gravity decreases by 0.16 percent from ‘Ain Dar to Haradh, a negligible amount compared to the changes in fluid density discussed above. As such, Earth’s gravity can be dismissed as a cause for the tilted OOWC in Ghawar.

It has been suggested that the rotation of the Earth affects large objects such as the Arab-D reservoir in Ghawar and may have caused the observed west-to-east OOWC tilt. However, the Earth has a counter-clockwise rotation by reference to the North Pole and, if effective, the inertial forces should cause a deeper OOWC on the west flank of the Ghawar field. On the contrary, field observations show that the contact is deeper on the eastern flank.

Assuming that Arab-D aquifer waters are vertically segregated according to salinity, we have shown numerically (see example in Appendix A) that a static equilibrium may account for the observed tilted OOWC along the main axis of the Ghawar field from south to north. This fits with the known pressure behavior of the field. However, the large west-to-east tilt in Central Haradh is more difficult to explain by the same calculations, as the depth to the equilibrium surface in the aquifer would be much greater. The shallow OOWC noted on Central Haradh’s west flank is localized and appears to be a physical singularity (Stenger, 1999). Fresh, recent aquifer waters have been sampled in this area and temperature variations seem to correlate with production-related pressure variations. In addition, the Wadi Nisah-Sahba fault zone (Figure 27) is near to all of these subsurface observation points. Although a strong correlation is no proof of cause, the authors believe that it is worth presenting the following interpretation that links all the available information.

• Wadi Sabha is clearly visible on satellite images (Figure 27). It is the easternmost extension of the larger Central Arabian Graben System, “an arcuate 560-km intraplate fault zone” (Al-Kadhi and Hancock, 1980) initially described by Powers et al. (1966). Most authors related the formation of the graben system to an extensional stress field (Vaslet et al., 1991). Weijermars (1998) however suggests that the Wadi Nisah-Sabha zone is the result of compressional forces creating a strike-slip fault system with about 8 km left-lateral displacement. According to Weijermars (1998) a substantial part of the movement on the Wadi Nisah-Sahba fault zone took place in the Pliocene-Quaternary. This was after the Arab-D trap had formed and presumably filled with hydrocarbons. A post-migration movement on the fault zone could explain how some oil remained trapped in oil-wet dolomitic stringers on the Haradh sector’s western flank well below the present OOWC.

• As discussed above, the possibility of fluid flow along the flanks of the Ghawar field is inconsistent with the field’s initial pressure data, the lack of aquifer support during production, and the conclusions reached by Dickey (1963) and Chiarelli (1973) concerning stagnant aquifers.

• The Pleistocene-aged aquifer water sampled from Central Haradh’s western flank was recently claimed (T.H. Keith, 1999, and S.W. Amos, 2002, personal communications) as proof of hydrodynamics with a recharge effect from the Arab-D outcrop near Riyadh, about 200 km to the west. We disagree with this interpretation for the following reasons:

  1. Chiarelli (1973) described the Arab-D aquifer in Saudi Arabia as rejecting system. With the exception of the western flank of Central Haradh, both the high chloride-calcium content and the isotopic signature of the Arab-D formation water around Ghawar indicate a stagnant aquifer.

  2. Pre-production static pressures do not show a pressure gradient across the Haradh area that would substantiate even localized hydrodynamics.

  3. Should hydrodynamics be active on Central Haradh’s western flank, the reservoir temperature would not drop when the pressure fell during the primary depletion phase.

  4. There is no obvious water outlet that would allow a hypothetical aquifer flow in the Arab-D.

  5. The Arab-D reservoir is a low-compressibility system and any significant rate of water movement would result in a large pressure gradient, as reproduced by the reservoir simulation model presented earlier (Figures 24a,b).

• The large W-E tilt seen on Central Haradh’s western flank is more likely to have been caused by limited vertical dumping from a shallower aquifer through fractures linked to the Wadi Sahba fault zone. This would explain the arrival of fresher, cooler water observed during the primary depletion of the reservoir. In this regard, it was noted that after the Haradh primary depletion and subsequent field shut-in in the early 1980s, only the pressure behavior of the west flank observation well HRDH-A could not be history-matched satisfactorily with the full-field reservoir simulation model (T.R. Pham, personal communication, 1999). Measured static pressures in HRDH-A showed a period of rapid pressure recovery until November 1988 when the pressure suddenly flattened out until the field was reopened in October 1990 (Figure 28). The well’s unique pressure behavior is interpreted as resulting from a degree of additional pressure support limited in time and space.

• Within a given reservoir, pressure-dependent communication across fault planes may start to leak fluids when a sufficient pressure difference—the pressure threshold—is applied to both sides of the fault (Zubari and Al-Awainati, 1997). In the case of Central Haradh’s western flank, the threshold pressure concept may be applied to a fault zone vertically linking the hydrostatic Jurassic Arab-D reservoir and a charged shallower aquifer such as the Cretaceous Biyadh (Figure 29). The value of the threshold pressure should be about 110 psi according to static pressures recorded when both the Biyadh aquifer and Arab-D reservoir were not under any production influence during the late 1980s (Figure 28). The Biyadh is the nearest shallower aquifer (Figure 29) having water salinity values almost identical to the salinity on the west flank of Central Haradh, and being affected by hydrodynamics. Natural dumping into the hydrostatic Arab-D reservoir from the shallower Biyadh would require a high potential to overcome the threshold and the Arab-D hydrostatic head.

• No water samples from the Biyadh aquifer that could confirm the match with the Arab-D water on the western flank of Central Haradh are available for isotopic analysis. This is because the Biyadh sandstones are unconsolidated and prone to instability. Hence there is an understandable reluctance to sample this particular reservoir while drilling. Although no pre-production pressure measurements were available for the Biyadh aquifer, the initial pressure of the immediately overlying Wasia aquifer (Figure 29) had been recorded near Central Haradh in 1953 before any local water production occurred. After converting the measured Wasia pressure of 1,115 psi at a depth of 1,166 ft to the Arab-D pressure datum (water gradient of 0.443 psi/ft), the Wasia initial pressure was 3,299 psi at 6,100 ft, compared with the original 3,216 psi of the Arab-D reservoir. Therefore, the Wasia aquifer was over-pressured by 83 psi compared to the Arab-D reservoir, a value that was consistent with the concept of a threshold pressure estimated at about 110 psi from later Biyadh pressure observations.

• The concept of the vertical dumping of water from a shallower aquifer into a deeper reservoir is not new. Khalaf (1989) proposed this for the Awali Arab-D field of Bahrain with the Cretaceous Shu’aiba aquifer dumping water into the Jurassic Arab-D reservoir. He described concomitant production problems such as plugging as a result of the deposition of asphaltenes due to reservoir cooling. In North Africa, Edmunds (1980) studied the geochemistry of the Miocene aquifer system in the Sirte Basin of onshore Libya. He identified “a striking feature *** a well-defined channel of fresh water, which can be traced in three dimensions across the main aquifer for some 130 km *** and represents recharge from a former wadi *** this groundwater, with an age of 7,800 +/- B.P., is chemically, isotopically and chronologically distinct from the older Pleistocene ground waters of the aquifer.” Edmunds’ description is similar to our current understanding of the impact at the Arab-D level of the Wadi Sahba fault, and his work shows that waters of different salinities do not mix easily in a porous medium.

• The possibility of a vertical communication between the Ghawar Arab-D reservoir and a shallower aquifer remains controversial as the Late Jurassic Hith Formation is believed to provide a very good seal to the Arab reservoirs (M.S. Ameen, 2001, personal communication). However, as discussed above, the Wadi Sabha trough is the eastward extension of the 95-km-long Nisah graben that is part of the Central Arabian Graben System (Figures 27 and 30). As such, the normal faults that bound the Sabha trough might have provided sufficient vertical offset to allow for partial vertical communication between the Arab-D and the Biyadh aquifers.

• The Mazalij and Ghazal Arab-D oil fields are located 80 km west from Haradh (Figure 1) on the northern and southern sides respectively, of the Wadi Sahba fault system. Recent 2-D seismic interpretation showed that normal faults have down-thrown the Arab-D reservoir by 400 ft south of Wadi Sahba (R. Geier and N. Al-Afaleg, personal communications, 2002; 2-D seismic interpretation by S. Dasgupta, 2001). The oil/water contact is also 400 ft deeper in the Ghazal Arab-D field in the southern fault compartment compared to the Mazalij Arab-D field in the north. This observation adds credence to the assumption that the tilted oil/water contact in Haradh is linked to a post-migration tectonic event along the Wadi Sahba trough. Further, open-hole logs in Mazalij wells show that all Arab reservoirs A, B, C and D are charged with oil in the wells closest to Wadi Sahba, whereas the northernmost wells are oil-bearing in the Arab-D reservoir only. This indicates that the Arab-D might have ‘leaked’ oil into shallower reservoirs in the vicinity of the Wadi Sahba fault. The observations made in the Mazalij and Ghazal Arab-D reservoirs show that surface observations related to the Wadi Sahba trough extend to the subsurface and have a significant impact on the Arab-D reservoir.

• In the nearby Awali oil field of Bahrain, Samahiji and Chaube (1987) described how tectonic breaching of the Hith Formation had allowed partial oil migration from the Jurassic Arab Formation to Cretaceous reservoirs. The conditions on the west flank of Central Haradh may be an intermediate situation in which the possibly leaking E-trending Wadi Sahba fault zone is intersected by the NW-trending Abu Jifan regional fault (Al-Husseini, 2000). This would explain why vertical communication could exist west from Haradh while the Arab-D trap was still fully charged with oil in Ghawar.

• It is the authors’ view that the concept of a vertical pressure threshold across the Wadi Sahba fault zone is reconciled with the observations. Specifically, it would explain how limited water dumping in the past may have created the west-to-east tilted OOWC in Central Haradh with a source point close to well HRDH-A. More likely, however, is vertical aquifer communication through the widespread and extensive karstic dissolution and collapse features (e.g. caves and sinkholes) known in Wadi Sabha and the eastward extension of the Central Arabian Graben (Figure 30). Collapse structures associated with the dissolution of Arab Formation evaporites were mapped by Vaslet et al. (1991) over an area of several thousand square kilometers (Figure 30). Two large karst features that have created 150-m-deep vertical fluid conduits through the Arab evaporite section, are present at the Hith Anhydrite reference section in a cave-like sinkhole at Dahl Hith and in another 3.5 km to the southeast (Vaslet et al., 1991). Other large sinkholes have been mapped and described, for example at ‘Ayn as-Dihl, Samah, and Umm Khisah near Al ‘Uyun, southwest of Al Kharj. The potential vertical extent of such karst features might be indicated by the collapse features mapped in the Dughum Member of the Biyadh Sandstone about 50 km east of Riyadh. Assuming these sinkholes, several square kilometer in size, are associated with the localized solution of the Arab evaporites in the subsurface, a vertical conduit of more than 600 m has been created.

• In the Ghawar subsurface, coherency analysis performed on 3-D seismic data, circular dissolution cavities that might be sinkholes were interpreted at several geological horizons. For instance, the shallower Cretaceous Ahmadi level shows sink holes apparently aligned on structural lineaments (Figures 31a,b). Therefore, karstic dissolution and collapse features could have a role in establishing a partial and vertical communication between the Arab-D and Biyadh aquifers.

The simplest mechanism that could account for the creation of the west-to-east tilted OOWC in Haradh is as follows:

• Tectonic activity along the Wadi Sahba fault zone allowed local and partial communication between the Biyadh and Arab-D aquifers on the western flank of Haradh.

• Due to its hydrodynamic regime, the higher-than-hydrostatic head of the Biyadh aquifer forced Biyadh water into the Arab-D on Haradh’s western flank and pushed the oil column up.

• Intrusion of water stopped after pressures came to equilibrium (less the threshold pressure effect), thus accounting for the static equilibrium and lack of pressure gradient observed today.

• Only when the Arab-D pressure dropped below the threshold pressure (as during the primary depletion phase in Haradh) did the dumping begin again.

In summary, regional evidence shows that the Hith anhydrite seal might be partially leaking to allow a remigration of oil from the Arab-D to shallower reservoirs, as in the Awali (Bahrain) and Mazalij oil fields. The geological and tectonic conditions under which the Hith anhydrite seal may have leaked are being reviewed at the basin scale and will form the basis of a future presentation.

The better understanding of the tilted OOWC in Ghawar has provided an answer to the problem identified by earlier workers in the field that, “Some type of static equilibrium is indicated by the high contents of dissolved solids in water, but its nature is not yet understood” (Aramco, 1959). Benefits become apparent in the ongoing development drilling of Central Haradh. The placement of producer and injector wells on the flanks of the Haradh sector using a minimum of delineation wells, the maintenance of reservoir pressure, and balanced production are all dependent on a correct appreciation of the tilted OOWC.

We thank the management of Saudi Aramco for their support and for permission to publish this paper. In particular, we thank N.G. Saleri, E.H. Bu-Hulaigah, and A.A. Al-Abdulkarim of Saudi Aramco for their support and encouragement. The authors are indebted to Moujahed Al-Husseini, Joerg Mattner and David Grainger of GeoArabia, Denis Mougenot, Abdulazeem Al-Towailib, and three anonymous referees, whose reviews allowed a major revision of the initial manuscript. The design and drafting of the final figures was by Gulf PetroLink.

The pressure exerted by a column of fluid may be expressed as follows:

Equation A-1 implies, however, that the fluid pressure gradient is constant over the area of application. Solving the pressure equilibrium equation for a structure filled with oil in its crest is equivalent to a traditional U-tube experiment, providing the assumptions of equation A-1 are fulfilled.

If the fluid pressure gradient is changing across a reservoir trap, equation A-1 has to be rewritten in an incremental form as:

The difference in altitude between two points is related to the horizontal distance as:

The cumulative fluid column pressure may be approximated as a finite summation over the structural axis, as follows:

Equation A-4 means that for a variable oil pressure gradient, the isobaric surfaces would be tilted in the oil zone according to the magnitude of the oil density change. If aquifer and oil densities are variables, the field equilibrium is found by calculating the depth in the aquifer (if any) at which the isobaric surfaces will return to the horizontal. A simple numerical application is shown below for a cross-section of a reservoir with variable oil and aquifer water densities (Figure A-1). Below the OWC on each side of the structure, the water pressure gradient is taken as constant. As shown in Table A-1, the combined weight of the oil and water column is in equilibrium at a depth of −7,691 ft on both sides of the cross-section with the selected OWC at −6,650 ft and −6,420 ft.

For the numerical application in Figure A-1 (constant absolute structural slope) and in the oil zone, Equation A-4 can be simplified between two successive points (x₁ and x₂) as follow:

Table A-1 shows the numerical application of static equilibrium at variable fluid densities. It is possible to derive and apply Equation A-5 to compute the pressure exerted by the oil column shown in the sixth column of Table A-1. The determination of the equilibrium depth is made with trivial additions to the water column weight until the pressure exerted by the total fluid column is identical on each side on the structure.

Bruno Stenger is a Senior Petroleum Engineer with Saudi Aramco working in the North Uthmaniyah Unit, ‘Udhailiyah Reservoir Management Division. He has a BSc in Geological Engineering (1989) from the École Nationale Superieure de Géologie, France and an MSc in Geomechanics (1990) from the Institut National Polytechnique de Lorraine. Before joining Saudi Aramco in October 2000, Bruno was the Chief Reservoir Engineer for H.O.T. Engineering, an oil and gas consulting company in Austria.

bruno.stenger@aramco.com

Tony Pham is a Petroleum Engineering Consultant with the Haradh and Harmaliyah Unit, ‘Udhailiyah Reservoir Management Division of Saudi Aramco. He has a BSc in Petroleum Engineering from Texas A&M University. Pham joined Saudi Aramco in 1982. He is involved in the development of drilling operations and planning for Central and South Haradh.

tony.pham@aramco.com

Nabeel Al-Afaleg is a Supervisor in the Haradh and Harmaliyah Unit, ‘Udhailiyah Reservoir Management Division. He has a BSc in Petroleum Engineering (1988) from King Fahd University of Petroleum and Minerals, Dhahran, and an MSc (1992) and a PhD (1996) in Petroleum Engineering from the University of Southern California. Nabeel Joined Saudi Aramco in October 1987.

nabeel.afaleg@aramco.com

Paul Lawrence is a Geological Specialist with the Southern Fields Characterization Division of Saudi Aramco. He has a BSc in Geology (1976) from Kent State University and an MSc in Geology/Geophysics (1978) from Wright State University, USA. Before joining Saudi Aramco in 1991, Paul was a Seismic Interpreter for Atlantic Richfield, Terra Resources, and Marathon Petroleum Tunisia. He is working on reservoir characterization and interpretation of 3-D seismic data in Ghawar field.

paul.lawrence@aramco.com