Numerical Simulation of a Reservoir Complex with a Tilting Gascap D5/D6 Sands - Utorogu Field, Nigeria

The Utorogu field is situated to the southeast of Warri, in the Western Operational Division of SPDC (Figure 1). The field has a total STOIIP of about 600 MMstb of which some 250 MMstb is contained in the D5/D6 reservoir complex. Production started in 1968 and by mid 1982 a total of 86 MMstb had been produced. Aquifer influx currently maintains the reservoir pressure close to initial. Production performance of individual wells and cased-hole logging has shown that the D5.2 reservoir has water influx from one side only.

Reservoir engineering analysis, with support from geology and petrophysics was carried out along the following lines:

• To obtain some feel for reservoir behaviour and aquifer strength, reservoir and well performance analysis together with material balance calculations were carried out.

• To analyse the gas-oil contact tilt and the resulting oil saturation distribution a numerical simulation was undertaken.

The study aimed at providing SPDC with 4 alternative production strategies and furthermore illustrates the application of both conventional and modern methods in the solution of a complex reservoir problem.

A geological study provided a model of the D5/D6 reservoirs. The structure of Utorogu is that of a rollover anticline bounded between two NW-SE trending growth faults. A small synthetic fault (estimated throw 40 ft) divides the field into a Main Block ‘X’ and a small Southern Block ‘S’ (Fig. 2). The reservoirs are Early Miocene in age and record successive periods of outbuilding and destruction of the early Niger Delta.

The D5.1 sand is a thin, rather irregular transgressive sand. There is some evidence of the D5.1 having the same fluid contacts as the D5.2 and D6.0. It is, however, unlikely that during production time the D5.1 will be influenced by cross-flow.

The D5.2 sand has an average thickness of 57 ft. The lower part comprises an upward coarsening, coastal barrier sand erosively overlain by a broad complex of channel sands with excellent reservoir quality. In the NW and locally elsewhere, thin flood plain shales form a vertical barrier. Finally 5-10 ft of coarse grained transgressive sands, deposited only in the extreme NE, completes the description of the D5.2 sand.

The D6.0 sand consists of a rather uniform coastal barrier overlain by generally fining upward transgressive sands. A thin marine shale separates the D5.2 and D6.0 sands. Smearing of this shale along fault-C is likely to hinder communication between the juxtaposed D6.0 in the Main Block and D5.2 in the Southern Block (Fig. 3).

The reservoir rock parameters were reevaluated to obtain a consistent set of basic data for the layering used in the model and are summarised in table 1. Porosity and initial water saturation were determined by establishing Φ S_h vs. R_T and Φ S_h vs. Φ relationships for each sand type in the four wells for which an FDC log was run in addition to the standard suit of resistivity, micro- and cased hole neutron logs. Values of ϕ and S_h could then be derived for the remaining wells.

Permeability was calculated using the equation k = 10⁵ Φ⁶ (1−S_h)⁻². This is the Schlumberger equation with an approximate adjustment for averaging on a log scale. Vertical permeability was set to 0.1 of the horizontal permeability in layers with net/gross sand ratio smaller than unity.

A common GOC at 8289 ftss and OWC at 8385 ftss were found for the D5.2 and D6.0 reservoirs. The D5.1 is believed to have the same fluid levels although the OWC is not penetrated. Cased hole CNL logs run in wells 5 and 16 in 1979, 10 years after the start of production, indicated considerable movement of the GOC when compared with b ase GNT logs. In well 16 the GOC has moved 10 ft deeper, whereas in well 5 gascap resaturation by oil has resulted in a 30 ft rise of the GOC (Fig. 4).

Capillary pressure is considered insignificant in view of sand quality. Moreover inspection of hydrocarbon saturation vs depth plots shows a negligible transition zone. Relative permeability data were chosen to match the individual well GOR's and watercuts.

It is assumed that all three reservoirs have the same fluid characteristics. PVT samples were collected from the D5.2 and D6.0 reservoirs, but neither at initial reservoir pressure, which is estimated at 3620 psig at a datum of 8310 ftss. The samples were subjected to a correction procedure to obtain a PVT set at initial conditions. Used data are depicted in figure 5.

Prior to further reservoir analysis the new maps were used to determine initial fluids in place (Table 2).

Observation of the reservoir pressure performance shows that water influx provides the dominant drive mechanism. Reservoir pressure at mid 1982 was approximately 3500 psig at datum, only 120 psi less than initial. Material balance analysis gives infinite aquifers and the aquifers contribute 90% and almost 100% to the drive mechanism in D5.2 and D6.0 respectively (Figs. 6, 7). Performance history of D5.1 is too short to allow a material balance calculation.

For the D5.2 and D6.0 bubble maps, depicting the areal distribution of withdrawals, were constructed (Figs. 8,9). The D5.2 bubble map reveals that water influx seems to come preferentially from the southeast as confirmed by the 1979 CNL logging. In the D6.0 reservoir this tendency is less pronounced but also present.

In summary the reservoir and well performance shows:

• The D5.2 reservoir has water influx mainly from the southeast, where wells produce water. Production from the entire oil rim has probably resulted in migration of the gascap to the northwest considering the high GOR production on that side of the reservoir.

• The D6.0 reservoir also has water influx, which is likely to come preferentially from the southeast.

Material balance calculations alone are inadequate to describe the reservoir fluid movements. Distribution of the oil in the D5.2 must be understood to be able to advise on further development plans. This three dimensional problem was therefore studied further by numerical simulation.

Geological reservoirs and fluid properties have been put together in a numerical simulation model. The use of a geological pre-processor greatly enhanced the derivation of grid block properties by interpolation from actual well data. A 7 layer subdivision of the D5.1, D5.2 and D6.0 reservoirs closely represents the distribution of barrier-bar, channel and transgressive sand types. Significant simulation data is listed in table 1 and the grid superimposed on the D5.2 structure map is shown in fig. 10.

Aquifers were attached to each reservoir. Aquifer representation is by analytical expression linked to the simulator as source terms. The D5.1 reservoir has influx only from the northwest, the good sand development being limited to the southeast. For the D5.2 reservoir only a southeast infinite aquifer is present. In the D6.0 reservoir the southeast aquifer is infinite and the northwest aquifer bounded.

Input to the history matching process is the oil rate for each well. The relative permeability curves were selected for GOR and watercut matching. At the same time however, pressure had also to be matched. The latter was determined by permeability level, aquifer size and location.

The final relative permeability curves showed the following characteristics:

• Residual oil saturation after a waterflood will be 20% (of pore volume). Both higher and lower values resulted in too little or too much movable oil and unsatisfactory breakthrough times.

• Residual oil saturation after a gasflood will be 5%. Again higher values could not match the breakthrough time. The critical gas saturation was 4%. As could be expected in high permeability reservoirs the low residual oil was considered indicative for good gas drainage.

• Oil curves are linear, water and gas curves have a Corey exponent of 5. The rock appears to be strongly water-wet (Ref. 2) and gravity drainage is favourable. Relative permeability sensitivity was such that moderate changes in endpoint relative permeabilities and shape did not give successful matches.

The balance between the relative permeability Corey exponent, permeability level and aquifer was difficult to obtain. After an initial set of sensitivity runs on these parameters numerous runs were required to obtain an adequate match. The permeability values derived with the Schlumberger equation were revised. The permeability changes consisted of a two-fold increase for the channel sands (D5.2) and a three-fold increase for the upper layer in D6.0. The Schlumberger equation used for permeability has been found to be inapplicable in the channel sands. The permeability increase in the upper D6.0 layer is due to the inclusion of part of the barrier sand which has a much higher permeability than the transgressive sand. Final matches for the D5.2 and D6.0 reservoirs are shown in Figs. 11a, b. Individual well matches for GOR and watercut were fair to good.

Important to the D5.2 reservoir behaviour is the gascap migration. In Fig. 12 a plot shows the 60% gas saturation contour in time. This illustrates that the gascap is shrinking primarily in the southeast with the prevailing water influx and the high GOR production. For two typical layers the areal oil saturation maps at mid 1982 are presented in Figs. 13a, b. Vertical oil saturation is presented in Fig. 14. In cross sectional figures (12, 14) the vertical exageration distorts the shape of the fluid contacts.

For the D5/D6 complex several future production strategies were analysed. Due to the shape of the D5.1 oil rim further activity was not considered.

These strategies are:

1. Continued depletion from existing wells and completions.

2. This strategy involves 6 recompletions and 1 infill well in the D5.2 to produce the migrated oil rim and 2 infill wells on the D6.0 reservoirs. In addition the western flank of D5.2 is closed-in until 1990 in an attempt to avoid resaturation losses.

3. As in case 2 with the western flank of D5.2 not closed-in, but including the drilling and recompletion activity.

4. As in case 3 with no infill wells drilled but including the recompletion activity.

The additional attainable oil recovery for these four alternatives is listed in table 3, rate versus time in Fig. 15.

Improved drainage can be obtained quickly (case 3) and medium to long term recovery is not enhanced by phasing (case 2). Avoiding infill drilling does not seriously deteriorate recovery for the D6.0. For the D5.2 sand infill drilling does noticeably improve recovery (case 4).

An important aspect in the usage of numerical simulation tools for reservoir analysis is the availability of adequate computing facilities. The computer configuration used consisted of:

• A Cray-1 computer for the numerical simulation calculations.

• A Univac 1100 for in- and output data manipulation.

• A VAX-11-780 as a driving machine for the GENISCO colour display facilities.

The hardware lay-out has been depicted in fig. 16. Available software consists of the numerical simulator and data processing packages.

The numerical simulator can handle 3 dimensional, 3-phase compressible reservoir problems and includes the effects of compressibility, gas solubility, gravity and capillary pressure. A typical Utorogu history match run (13 years) took just over 10 minutes of CPU time on the Cray-1 computer at a cost of some ESTG200. Standard output is on the printer. The general disadvantages of this type of output are that it is bulky and difficult to interpret. The developed conversational plotting package allows the plotting of saturation and pressure maps, and performance plots for wells, reservoirs, layers and groups of these (Figs. 12,13). Reproduction on a VDU screen allows instantaneous inspection of key parameters after a simulation. The use of such a package is important in view of the short turnaround time of a Cray-1 computer. Full advantage can be taken of its speed by reducing the time of the diagnostic process after a run. Instructive potential is an extra benefit from the plotting package.

Capabilities of the Genisco colour display facilities are primarily in showing the movement of fluids in time.

Reservoir development analysis has lead to the following conclusions.

1. Further development of D5.1 can only be undertaken as a secondary target. The oil rim is too small for dedicated wells.

2. For the D5.2 further development must consider recompletions. Infill drilling improves recovery noticeably. Additional recovery is some 9% of STOIIP for the recompletions. Infill drilling adds about 2% of STOIIP for the first well.

3. The D6.0 reservoir is fully developed. Existing wells can produce the remaining movable oil in an acceptable timespan.

The methods applied illustrate:

1. A combination of conventional and modern tools allows a reservoir engineer in 1983 to obtain a good insight into the reservoir drive mechanisms. It must be emphasised that conventional tools are primarily for 1 dimensional problems whereas modern tools allow 3 dimensional, multi-phase problems to be solved.

2. An advanced computer system, well equipped with the necessary software, greatly enhances the reservoir analysis process.

3. The time required for a study and diagnostic potential is improved with the introduction of adequate pre- and post processing software to a numerical simulation package.

ABBREVIATIONS
AbbreviationExpansion 
• CPU

  Central processing unit

 
• FGIIP

  Free gas initially in place

 
• GOC

  Gas-Oil contact

 
• GOR

  Gas/Oil ratio

 
• OWC

  Oil-Water contact

 
• SGIIP

  Solution gas initially in place

 
• SPDC

  Shell Petroleum Development Co. of Nigeria Ltd.

 
• STOIIP

  Stock tank oil initially in place

 
• VDU

  Visual Display Unit

SYMBOLS
AbbreviationExpansion 
• B_g

  Gas formation volume factor

 
• B_o

  Oil formation volume factor

 
• k

  Permeability

 
• m

  Ratio gascap/oil column hydrocarbon pore volume

 
• S_h

  Hydrocarbon saturation

 
• R_s

  Solution gas/oil ratio

 
• R_T

  Total resistivity

 
• µ_g

  gas viscosity

 
• µ_o

  Oil viscosity

 
• Ф

  Porosity

The Authors want to acknowledge the stimulating comments and contributions of H. Niko and the work performed by J. Holloway and G. Whibley. Shell Petroleum Development Company of Nigeria Ltd. and Shell Internationale Petroleum Maatschappij B.V. are thanked for their cooperation in preparing the paper.