Summary Reservoir engineers face a large number of possible development plans for major offshore oil fields. Some of the factors affecting development include product-processing capacities, drilling-rig availability, extended use of existing platforms, enhanced-recovery options, and subsidence-avoidance requirements. Reservoir simulation is an accepted technique for estimating production from oil and gas fields. Each possible development scenario is simulated separately to obtain future oil- and gas-production estimates from wells included in the scenario. Simulation results are analyzed for each run, and the optimal plan is selected. Enumerating all the possible drilling schedules is a difficult task, and, even with clusters of advanced scientific workstations, the computational requirements of simulation can be overwhelming. This paper describes the application of linear mixed integer programming (MIP) to the combinatorial problem of generating the optimal redevelopment scenario for the Ekofisk field in the Norwegian sector of the North Sea. The MIP model determines the optimal redevelopment program and drilling schedule while maintaining operationally feasible product-processing requirements, drilling-rig use, and well/platform relationships. Optimality is based on the net present value (NPV) of the estimated production forecasts for the wells selected. Total recovered reserves is an alternative optimization criterion. Reservoir simulations generated the estimated production forecasts used in iterative runs of the optimized model. We discuss success of the approach and extensions. Introduction The field-redevelopment project presented enormous challenges to optimizing value while scheduling drilling of more than 60 new wells and timing permanent platform shutdowns within physical facilities constraints. In addition, because many of the wells would be completed in the same reservoir, the order of drilling affected the production rates from subsequent wells. The large number of possible development schedules and production profiles for each well indicated that a computer application would consider the possibilities and optimize the planning faster than more traditional methods. We chose linear programming because of its effectiveness in solving scheduling problems in other industries. Ekofisk Field History The Ekofisk field and surrounding Phillips Norway group fields, also referred to as the greater Ekofisk area fields, are in the southern part of the Norwegian sector of the North Sea. The Ekofisk field produces from a Maastrichtian/Danian chalk reservoir of the Tertiary/Cretaceous periods. The reservoir structure is an elliptical anticline, 6.6 miles long and 3.2 miles wide. The reservoir thickness is between 300 and 1,000 ft, with the top of the reservoir located at 9,500 ft subsea true vertical depth. The reservoir conditions include high pressure, high temperature, and high porosity with low rock-matrix permeability and extensive natural fracturing. Primary oil production from Ekofisk field began in 1971, with all gas reinjected. In 1977, gas sales began; however, gas injection continued for contractually obligated swing-gas volumes in excess of demand. Full-field secondary waterflood operations began in 1987.1 At year-end 1996, 65 producing wells, 36 water-injection wells, and three gas-injection wells were located on four production platforms and two water-injection platforms in the Ekofisk field. Processing, transportation, and accommodation facilities were located on eight additional platforms at the Ekofisk center. Seven surrounding Phillips Norway group fields produce from 77 producing wells located on nine production and processing platforms. Total production for Ekofisk and the surrounding fields at year-end 1996 was approximately 320,000 BOPD and 830 MMscf/D. Oil- and gas-separation and -transportation facilities are centrally located on the Ekofisk complex at Ekofisk field. The Ekofisk II Redevelopment Project is designed to replace the oil- and gas-production and -processing capabilities of the existing Ekofisk complex. This requirement grew out of the high-operating and -maintenance expenses associated with the existing facilities. Other factors of significance were the effects of seafloor subsidence and changing safety regulations. A significant aspect of the Ekofisk field has been reservoir compaction that resulted in seabed subsidence during the areal extent of the reservoir. After 25 years of production, the cumulative subsidence in the center of the field is more than 21 ft. The Ekofisk II Redevelopment Project addresses the economic, maintenance, and safety factors and maintains the economic viability of Ekofisk field and the surrounding Phillips Norway group fields.

A case study is presented showing how reservoir characterization with geostatistical modeling was used to develop the reservoir flow simulation model for Bayu-Undan Field. The modeling was tailored to both reservoir type and the selected development scenario. Bayu-Undan Field is currently under development as a gas recycle project with future gas export. The reservoir is made up of both fluvial and marine sedimentary sequences. The need to predict fluid flow for both recycle and depletion scenarios in the presence of these reservoir types requires an understanding of reservoir heterogeneity and a method to preserve it in an upscaled flow simulation model. The major control on reservoir character is facies type which controls grain size, sedimentary structure and degree of bioturbation. Facies were determined from log motif and core description. Different statistical techniques were chosen to build facies models for the continuous marine zones and the heterolithic marine/fluvial zones. Numerous facies realizations were constructed to evaluate the possible range of outcomes and uncertainties. Porosity was distributed either deterministically using a weighted average algorithm or statistically using variograms. Permeability was distributed using a porosity-based transform. A number of permeability upscaling algorithms were evaluated to ensure reservoir heterogeneity was preserved when upscaling from the geocellular model to the flow simulation model. As a result, separate algorithms were chosen for horizontal permeability in marine and fluvial units, and for vertical permeability. Use of these upscaling techniques resulted in very good preservation of gas in place by permeability class, shown by comparing the geocellular model with the upscaled model. The modeling captured and preserved the reservoir flow characteristics critical to performance predictions for the gas recycle project. Introduction The Bayu-Undan Field is located in the Zone of Cooperation A (ZOCA) in the Timor Sea 280 miles northwest of Darwin, Australia. It is approximately mid-way between Australia and the island of Timor, Figure 1. The field was discovered in 1995 and is a retrograde gas-condensate reservoir1. Reserves are 400 MM bbls of associated liquid hydrocarbons with potential gas sales of 3.4 tcf. The reservoir rock is low porosity sandstone, 8–15%, with relatively high permeability, up to 2000 md. In addition to the discovery well, 10 appraisal wells have been drilled. Extensive open hole logging has taken place and approximately 3500 ft of reservoir section has been cored. A total of 15 drill stem tests (DST) were taken on the discovery and appraisal wells. Information from the log analyses, detailed core descriptions, DST analyses and further petrologic studies form a huge database describing the reservoir. A gas recycle liquids recovery project is currently under development but the option for future gas export remains open. Due to its offshore location, field development costs will be high. To control and reduce development risks and uncertainties, reservoir performance modeling needed to be of the highest practical accuracy. The extensive database and access to state of the art modeling software provided the tools for this reservoir characterization project.

SPE Members Abstract A major field study was conducted to determine future operating strategy for Ekofisk Field. Historical and predicted performance of various IOR processes were evaluated and an IOR process was selected. The evaluation was made in the context of ongoing Upper Ekofisk Formation gas injection and water injection into the underlying Lower Ekofisk and Tor formations. Optimization of the recommended waterflood and studies of other reservoir management concerns are described. The reservoir management strategy was developed to integrate the most economical improved oil recovery process with a strategy to minimize future subsidence. Introduction Ekofisk Field is a mature oil field currently being waterflooded for improved recovery in the lower two thirds of the reservoir. Starting in 1991, a major field study was undertaken to determine the future operating strategy for the field. Historical performance and predictions for improved oil recovery (IOR) processes were evaluated and an IOR process selected. The evaluation considered ongoing hydrocarbon gas injection and pilot water injection in the Upper Ekofisk as well as ongoing water injection into the underlying Lower Ekofisk and Tor formations. In addition, optimization of field wide IOR processes was carried out to maximize recovery while managing sea floor subsidence. In late 1992, the results of the study were presented as an Extended Field Study (EFS) to the Norwegian Petroleum Directorate. Background Ekofisk Field is located in the Norwegian sector of the North Sea and currently produces approximately 150,000 BOPD and 680,000 MSCFD from 70 wells. Current water injection into the lower two thirds of the field is 500,000 BWPD into 35 wells. The structure is a large elliptical anticline, 6.6 miles long and 3.2 miles wide, with the top of the structure at 9500 feet sub-sea true vertical depth (TVD). Structural relief is significant only because of the size of the field; the angle of dip is very low, averaging less than 4 degrees. The reservoir varies from 300 to 1000 feet in thickness and can be subdivided into 3 major geologic zones: the Upper Ekofisk, Lower Ekofisk and Tor formations. The Tor Formation is composed of porous chalk of Maastrichtian (U. Cretaceous) Age. Overlying the Tor Formation is a low porosity impermeable zone which comprises the base of the Danian Age (Paleocene) Lower Ekofisk Formation. The remainder of the Lower Ekotisk Formation consists of reworked Cretaceous chalk and has similar reservoir characteristics to the Tor Formation. The Upper Ekofisk Formation was deposited in a lower energy environment and has generally lower porosity and permeability. Each of the major intervals contains approximately one third of the original oil in place. Ekofisk Field is a low permeability fractured chalk with matrix permeabilities ranging from 0.1 to 10 md. Effective permeabilities are 2 to 50 times the chalk matrix permeability. Vertical permeability ranges from 0.1 to less than 0.01 times that of the horizontal effective permeability. The chalk is a soft, low strength sediment with reservoir porosities ranging between 30% and 45%.

The Ekofisk Field in the Norwegian North Sea is currently being redeveloped with the installation of a new processing platform and a new 50 slot wellhead platform. The new processing facilities are scheduled to be installed in 1998 and will serve as a central processing point for the Ekofisk field and all of the outlying fields in the Greater Ekofisk Area. Drilling operations on the new well head platform will commence in 1996 and will result in the drilling of up to 50 wells over the next eight years. In order to maximize the value of this major redevelopment, a project was initiated to optimize the development strategy for Ekofisk and the outlying fields. This paper describes the tools developed during this project. Two different techniques of coupling linear programming and reservoir simulation were developed independently using iterative approaches to obtain solutions. The results of the two approaches yielded very similar solutions, which provides a high degree of confidence in the solution techniques. Introduction The redevelopment of the Ekofisk field, also known as the Ekofisk II project, has a significant impact on both Ekofisk and the outlying fields in the Greater Ekofisk Area. The new 2/4 J Processing Platform will be installed in 1998 and will consolidate all Greater Ekofisk Area Oil and Gas processing on one platform. The new processing platform will accommodate production from the Ekofisk Eldfisk, Embla and Tor fields, as well as any future developments. The new 50 slot 2/4 X Wellhead Platform will be installed in 1996 and used to redevelop the Ekofisk Field, eventually replacing the existing 2/4 Alpha and 2/4 Bravo Platforms. Early in the stages of the Ekofisk II project, it was decided to design the new processing facilities to plateau production for the first few years. While plateauing production generally provides the most attractive economic alternative, it significantly complicates the development planning for a field. In the case of the Ekofisk II project, a large number of economically viable opportunities existed to increase production and a procedure was required to optimize scheduling of these projects. Table 1 summarizes the number of projects that were under consideration. The rate at which these projects could be developed was governed by two major constraints: processing capacity and rig availability. The new Ekofisk II processing facilities limited gas processing to 789 MMSCFD and oil to 274 MBOPD. Drilling constraints varied by platform. Fixed platform rigs capable of drilling six wells per year existed on Ekofisk X and Eldfisk Bravo platforms. One jackup rig was also available, and was capable of drilling six well per year on Ekofisk X, Eldfisk Alpha, or Tor Platforms. The jackup was also capable of drilling on Embla, but was limited to four wells per year due to the longer drilling times for the deeper, higher pressured Embla reservoir. No drilling was premised on Ekofisk Alpha, Ekofisk Bravo or Ekofisk Charlie platforms. Table 2 summarizes the drilling constraints for the Ekofisk II project. Within these constraints, a number of different development strategies were possible. Four of the most significant options that required evaluation were as follows:Accelerating drilling on outlying fields. Excess processing capacity existed prior to start up of the new process facilities in 1998. Opportunities existed to increase production on the smaller outlying fields prior to 1998, but the long term production impact of these projects would reduce the processing capacity available after start up of the new facilities. The value of this accelerated production had to be weighed against the potential impact of delaying higher value Ekofisk X Platform wells until after the plateau period. P. 507

The Bayu-Undan Field - located offshore in the Timor Gap Zone of cooperation Area A (ZOCA) between Australia and Indonesia is a significant "world class" gas-condensate resource. A group led by Phillips Petroleum Company discovered the field in 1995 with the drilling of the Bayu #1. A group led by BHP Petroleum confirmed this discovery as a large accumulation with the drilling of the Undan #1. These two wells are 6 miles apart. This paper will present the results of the appraisal and conceptual development planning of the field to date. Ten wells have been drilled into this normally pressured sandstone reservoir with consistent results as to the fluid composition and contacts. The reservoir contains a relatively lean gas-condensate whose tested condensate to gas ratio (CGR) is 60 bbls/MMscf. Laboratory testing indicates 2.4% liquid dropout. The gas-water contact (GWC) in all wells is located at 10197 ft subsea. Although large faults creating the possibility for several fault blocks are apparent in all of the seismic interpretations, the faults have not led to differing fluid compositions or contacts. The reservoir is comprised of a series of sandstone and shale intervals. In general, the upper part of the reservoir contains sands of lower net-to-gross and lower quality, while the lower part of the reservoir contains sands of higher quality with higher net-to-gross values. Core and drill stem testing has shown large variability in reservoir quality. Permeabilities from both core and drill stem tests range from less than 1 millidarcy to over 1 darcy. Introduction The Bayu-Undan Field was one of the most significant discoveries in 1995 in terms of total reserves of oil and gas. The field, located in the offshore Timor Gap Zone of cooperation Area A between Australia and Indonesia, is 280 miles northwest of Darwin, Australia (Fig. 1). The nearest landfall is approximately 130 miles north on Timor Island, Indonesia. The major part of the field lies in water depth between 230 to 300 feet. The field straddles two adjacent Production Sharing Contracts - ZOCA 91–12 operated by a group led by BHP Petroleum and designated Undan and ZOCA 91–13 operated by a group led by Phillips Petroleum Company and designated Bayu (Fig. 2). The appraisal and development planning of the field is being carried out to commercialize the field at the earliest opportunity. The discovery well, Bayu #1, drilled by Phillips Petroleum Company intersected 575 feet of hydrocarbon column and the confirmation well, Undan #1, drilled by BHP Petroleum intersected 410 feet of hydrocarbon column. Both wells have the same GWC of 10197 feet subsea. To date, eight additional wells have been drilled, thus accomplishing a successful appraisal campaign. In the last twelve months, extensive subsurface and feasibility studies to finalize and optimize the development plan have been undertaken. Geology The Bayu-Undan Field is part of the greater Flamingo High within the northern Bonaparte Basin. The east-west trending Flamingo High is surrounded by north-northwest trending Flamingo and Sahul syncline and east-west trending Malita Graben, which may have been the source area for the Bayu-Undan hydrocarbons. A thick sequence ranging in age from Triassic to Recent has been intersected in the area. The detailed regional geology has been documented by several authors. The description of the geology of the Bayu-Undan area is given in a paper by Brooks, et al. The Bayu-Undan structure is a broad east-west trending horst with a number of culminations set up by internal faults (Fig. 2). The field covers an area approximately 18 miles in length and 9 miles in width. P. 401^