Application of Open-Architecture Computing Environment in Reservoir Simulation

Abstract

Summary Open architecture has reshaped the computer marketplace and has significantly affected the development, maintenance, and application of reservoir simulation technology. We have moved rapidly from mainframe computers to distributed applications on workstations. Here, we discuss four open-systems initiatives (communication, language, operating system, and graphics) that have particularly influenced reservoir simulation at Mobil R and D Corp. Over the last 2 decades, our simulators have grown from 50,000-line single-well programs to greater than 1.5-million lines of code in the current field simulator. Advances in open systems have enabled us to manage this explosive growth on a network of workstations. We have been able to standardize around the UNIX operating system and C and FORTRAN languages. X-Window MOTIF-based interactive graphics have been a key element in migrating the simulator to a variety of computers. Although these open-architecture initiatives have had a significant and positive impact on reservoir simulation at Mobil, we still are faced with certain disadvantages. This paper discusses how we overcame these disadvantages and the challenges we face in the future. Introduction "Open systems" has become one of the most overused computer buzz words, yet that concept is credited with reshaping the marketplace. This paper describes four open initiatives as interpreted at Mobil and presents a metric to quantify their impact on the development, maintenance, and application of the company's reservoir simulators. For the purposes of this paper, a reservoir simulator is simply a computer program whose heart is a computationally intensive numerical model. The program processes the physical description of the reservoir (e.g., the type of fluid, pressure, saturation, temperature, and geology). It also processes the proposed development strategy in the form of operating constraints. Using this information, it predicts the effectiveness of different production strategies and is used to optimize gas and oil recovery from petroleum reservoirs. During the last 2 decades, Mobil has developed four reservoir simulators. Fig. 1 shows that the number of lines of code in each program increased dramatically as more complex engineering features were incorporated. The first of these programs, WELCOS, was released more than 20 years ago. That black-oil simulator could model the behavior of only a single well and was used extensively for water- and gas-coning studies. Written in FORTRAN, it consisted of less than 50,000 statements. By 1975, Mobil had developed ALPURS. It was also a black-oil model, but ALPURS had the power to model all the wells in a field. ALPURS contained sophisticated well controls that enabled it to model complex production strategies. ALPURS also provided limited, black-and-white plotting and contouring capabilities. Because of its added engineering capabilities, ALPURS was six times the size of WELCOS, or about 300,000 lines. In the mid-1980's, Mobil developed a fullfield, compositional reservoir model, COSMOS. It was about one-third larger than ALPURS, with more than 400,000 statements. PEGASUS is a multipurpose, compositional, black-oil simulator capable of modeling a broad range of reservoirs with complex geology and fluid behavior, each with different production strategies. It integrates various modeling capabilities, such as a sophisticated well-management scheme for full-field studies and accurate predictions of horizontal-well performance. PEGASUS is used to study gas-condensate, volatile-oil, faulted, and fractured reservoirs. It also has gas-storage applications. It is 30 times the size of WELCOS and contains more than 1,500,000 statements. The interactive color graphics feature for interpretation of simulation results represents about 40% of the code.