Incorporating Uncertainties in Well-Count Optimization With Experimental Design for the Deepwater Agbami Field

Abstract

This paper describes a methodology for incorporating uncertainties in theoptimization of well count for the deepwater Agbami Field development. The lackof substantial reservoir description data is common in many deepwaterdiscoveries. Therefore, the development plan must be optimized and proven to berobust for a wide range of uncertainties. In the Agbami project, the design ofexperiments or experimental design (ED) technique was incorporated to optimizethe well count across a wide range of subsurface uncertainties. The lack of substantial reservoir description data is common for manydeepwater discoveries. In the Agbami project, the uncertainty in oil in placewas significant (greater than a factor of two). This uncertainty was capturedin a range of earth (geologic) models. Additional uncertainty variables, including permeability, fault seals, and injection conformance, were studiedconcurrently. Multiple well count development plans (high, mid, and low) weredeveloped and used as a variable in ED. The ED technique allowed multiple wellcounts to be quickly tested against multiple geologic models. With the netpresent value (NPV) calculated for each case, not only was the well count forthe overall highest NPV determined, but discrete testing of each geologic modeldetermined the optimum well count for each model. The process allowed testingthe robustness of any well count versus any uncertainty (or set ofuncertainties). A method was demonstrated quantifying the difference between perfect andimperfect knowledge of the reservoir description (geologic model) as itpertains to well locations. Introduction The Agbami structure is a northwest/southeast trending four-way closureanticline, and is located on the Niger delta front approximately 65 milesoff-shore Nigeria in the Gulf of Guinea (see map in Fig. 1). Thestructure spans an area of 45,000 acres at spill point and is located in 4800ft of water. The Agbami No. 1 discovery well was drilled in late 1998. Theappraisal program was completed in 2001 and included five wells and onesidetrack drilled on the structure with each encountering oil pay. These fivewells and a sidetrack penetrated an average of approximately 350 ft of oil. In this phase (Phase 3) of the development process, the key objectives areto construct a field development plan and to obtain sanctioning. With drillingdepths of up to 10,000 ft below mudline in 4800 ft of water, well costs atAgbami will be at the high end of typical deepwater costs. Therefore, animportant optimization parameter in the field development is the total wellcount. Agbami is typical of many deepwater developments in that the seismic is lessthan perfect and the appraisal well data are sparse relative to the areacoverage. Therefore, subsurface uncertainty is high. In fact, the 5% probableoil in place is more than two times the oil in place at the 95% probability. Asa result, the development process is challenged with determining the optimumwell count for the field development across the wide range of subsurfaceuncertainty. Several key development decisions were determined in the previous phase(Phase 2) of the development process. These decisions were taken as givens inthis study and are listed as follows:Recommended pressure maintenance scheme and gas disposition strategy for the17 million-year (MY) units is a combination of crestal gas injection withperipheral water injection.Recommended pressure maintenance scheme and gas disposition strategy for the14MY/16MY units are crestal gas injection only.Facility design capacity recommendations are:250,000 stock-tank bbl per day (STB/D) oil450,000 thousand cubic ft per day (Mcf/D) gas production250,000 STB/D water production450,000 STB/D liquid production450,000 STB/D water injection

Reservoir Forecast Optimism – Impact of Geostatistics, Reservoir Modeling, Heterogeneity, and Uncertainty

“…Note that many of the models generated with very small cells (e.g. There are many excellent papers and textbooks that discuss both uncertainty assessment and experimental design so the details of these subjects will not be treated in this paper (see, for example, (Freiedman et al, 2003;White and Royer, 2003, White et al, 2001, Montgomery, 2001, Peng and Gupta, 2003, Sanhi , 2003, Kabir et al, 2004, Narahara et al, 2004, Anderson and Whitcomb, 2000, Barratine, 1999, Mason et al, 2003. Figure 3 shows a high level view of the typical modeling and forecasting workflow used in the 1990s.…”

“…Recent papers addressing aspects of well placement optimization include Baris and Horne (2001), Ozdogan and Horne (2004), Badru and Kabir (2003), Wand et al (2007), Emerick et al (2009). Recent papers addressing aspects of well placement optimization include Baris and Horne (2001), Ozdogan and Horne (2004), Badru and Kabir (2003), Wand et al (2007), Emerick et al (2009).…”

Tahiti: Development Strategy Assessment Using Design of Experiments and Response Surface Methods

“…[10][11][12][13][14][15][16][17][18][19][20][21] Integrated workflows and automatic processes were presented in a few publications but they are not widely spread in the industry. 22,23 A lot of opportunities remain to integrate the design of experiments and response model framework in a unique software package.…”

A Comparison Study on Experimental Design and Response Surface Methodologies

“…For each case, all cdfs are compared to the reference obtained from the results of the exhaustive runs. The deviation D from this target cdf is estimated by computing the expected value of the relative difference between distributions as shown in Equation 12. The results summarizing the deviations from the reference are presented in Figures 7 to 9.…”

“…The above workflow, sometimes with variations, has been extensively applied on various engineering studies. [1][2][3][4][5][6][7][8][9][10][11][12][13][14] In this study we aim to perform a comparison on the experimental design methodologies and the response surfaces which are used during steps 3 and 4 (and hence steps 8 and 9) on a number of synthetic and real field examples. We also explore the adequacy of the Plackett-Burman design in terms of its capability to estimate the most influential parameters on the system.…”