No abstract

Mature fields allow for a novel approach to generate forecasts for infill drilling, borrowing techniques developed in marketing. Fields with a long production history usually have an abundance of both basic data and production data, which can be linked and utilised to obtain reliable forecasts. This takes away the need to build a time-consuming, full field simulation model. The forecasting method is referred to as Data Driven Predictive Analysis. This document focuses on an application of Data Driven Predictive Analysis on a mature oil field in the Sultanate of Oman. A Data Mining Tool is used to find correlations between well performance and reservoir data. These correlations are investigated in order to construct the optimal Bayesian network to describe the link between basic field data and infill well performance. The relevant data from the Data Mining Tool are used to train the Bayesian network. The trained network predicts the performance of new infill wells based on their expected properties, which are derived from a static model. A new feature presented in this paper is the generation of a 2D grid of infill well forecasts including uncertainty ranges directly from the static model, by importing geological property grids into the Bayesian network. Water coning is taken into account by including perforation standoff as a function of distance around producing wells, as derived from radial single well models. TX 75083-3836, U.S.A., fax +1-972-952-9435

Shell and PetroChina have been successfully developing the Main Changbei P1S2 QA Reservoir using the advanced technology of openhole dual lateral horizontal wells since 2005. The scope of this project has been expanded to further develop the Changbei Main Reservoir and also to appraise and develop the Unconventional tight gas Reservoirs. The Unconventional tight gas Reservoirs have not been commercially developed to date and are geologically very different from the Main Reservoir. The team employed front end loading and decision based modelling to manage the complexities of essentially developing a green field within a brown field, and unconventional microDarcy reservoirs above and below a more conventional milliDarcy reservoir. The field was divided into type areas based on reservoir quality and development maturity. The Main Reservoir modelling approach is conventional but the Unconventional Reservoirs will require single-well multi-reservoir box models to facilitate probabilistic uncertainty work flows and type curve generation. Synergies between the Main and the Unconventional Reservoirs have also been considered.

No abstract