Management and portfolio evaluations is one of the crucial tasks that accurately account for valuating a petroleum asset. Usually, variations in input parameters and uncertainty on the evaluation of oil assets affect those processes, as it has been occurring with the price of oil in the last year.

This paper presents a new systematic process for the evaluation of portfolios of oil and gas fields where the performance and economic value of an entire portfolio decrease rapidly. The automated cash flow-curve analysis tool presented here uses an event detection algorithm combined with the design of quantile regression technique to provide a robust probabilistic estimate of future PDP (proved-developed-producing) reserves on a well-by-well basis. Individual well behaviors then aggregate stochastically to provide expected field and portfolio declines, with uncertainty ranges. Future well trends are estimated using probabilistic type-curves computed by data mining algorithms with a high-level of granularity.

Most discussions and publications to date have centered on the methods to perform portfolio optimization. However, very little emphasis was put on using analytical and data mining methods to evaluate faster the status of a given portfolio. This paper will focus on the use of data mining related methods in analyzing oil assets/engineering data, identifying correlations, development of industry-based algorithms and in the determination of relationships that influence root cause and consequence of failure in mature fields.

This paper proposes a new approach for mature fields. It includes models with technical Data Mining reserves estimation (either applied for volumetric calculations or performance data based in technical of Extracting hidden knowledge), an scenario matrix to account for the risk expressed in expectation curves, an estimation of the "Pseudo-limiting" risk and value terms, and defining a hierarchy of portfolio with data mining technique. Furthermore, it proposes a methodology of life-cycle assessment and surveillance of reserves estimation by integrating of the "Pseudo-limiting" risk and the ratio expected value to capital investment under the concepts of risk management.

The developed models showed good performance with minimal prediction errors. These results are promising, lending credence to the application of computational Intelligence for even more complex reservoir systems. They should also boost confidence in the use of advanced well structures for field applications.

The consideration of such needs bring up the following purposes:

Extracting hidden knowledge or information not trivial of dataset to be used in making decision.

Discovery of unknown models [1] [2] in order to discover meaningful patterns and rules [3].

This content is only available via PDF.
You can access this article if you purchase or spend a download.