Simulated Expert Interpretation of Regional Data to Predict Drilling Risk

Abstract

Abstract Incomplete or sparse information contributes to high levels of risk for oil exploration and development. To more accurately and consistently predict drilling risk, a degree of automation of data analysis may be helpful. "expert systems" developed and used in several disciplines and industries, have demonstrated beneficial results in modeling the decision making process of human experts. A state-of-the-art "expert" exploration tool using computerized multidisciplinary databases, expert developed "rules", and regional data maps — generated using artificial neural networks, has been developed. The system employs a web interface for users to select prospect(s) of interest and to allow data review or addition, and includes security to maintain the sanctity of proprietary information. Two types of rules are applied to the data. Heuristic rules are generated directly from engineering, geophysical and geological databases. Expert rules are developed through interviews with successful prospectors. Rules are applied in four categories: Regional Indications, Trap Assessment, Formation Assessment, and Oil Price. Some users may elect to not factor in certain aspects, or to use their own values. Each of the sub systems assigns a numerical score based on the answers to individual "expert" questions. Results are then combined to form an overall risk assessment associated with the selected prospect(s). These scores can be derived from "crisp" mathematical computations, "fuzzy" analysis, or a combination of the two. This expert system can help companies of all sizes, to more efficiently evaluate prospect quality, and thus more rapidly eliminate poor prospects and generate new production from favorable prospects. The initial system has been initially tuned for the Brushy Canyon formation, Delaware basin, New Mexico. Introduction As smaller companies and tighter exploration budgets increasingly predominate in the onshore U.S., it becomes increasingly important to both warehouse technical knowledge and increase the cost efficiency of finding both new fields and new wells in existing fields. Expert systems are an artificial intelligence tool which stores expert opinions and methodologies of analysis. They have been successfully implemented in the medical field to make diagnoses given lists and degrees of symptoms1. Expert systems have also been implemented in the oil industry, but generally only to examine narrow problems with a very small focus of expert knowledge required2,3,4. This paper describes a new expert system for oil prospecting which is currently under final development. This Fuzzy Expert Exploration Tool (FEE Tool) will eventually be generalized so that users in any part of the world will be able to add their own knowledge and data and make rapid evaluations of a large number of potential drilling sites in a systematic and consistent manner via the internet. A case study of the Lower Brushy Canyon formation of the Delaware basin, New Mexico is being used to verify the software and test the potential of the FEE Tool. In the first four years of the FEE Tool Project, an immense amount of data on the Delaware Basin has been accumulated, including data on geology, structure, production, regional information such as gravity, and local data such as well logs. This data, organized and cataloged into several online databases, is available for the FEE Tool and users as needed and as appropriate in analyzing production potential. In addition, a preliminary map of predicted production for the Delaware basin5 has been generated using a neural network with regional data as inputs and production potential as the output (average barrels per month of oil expected for the first twelve months).