| Popis: |
Geosteering technology has played a key role in enhancing hydrocarbon production and recovery in many reservoirs throughout the world. Traditionally, Logging While Drilling (LWD) measurements are used to determine petrophysical parameters, such as porosity and hydrocarbon saturation in order to geosteer wells in the reservoir. However, in many complex carbonate reservoirs, fluid flow characteristics are generally difficult to predict and the most porous intervals are not always the best reservoirs as intervals of equivalent porosity can exhibit large variations in permeability. In addition, changes in depositional and digenetic environments provide for a complex geology that poses challenges in well placement. A new technique has been developed that utilizes LWD data and neural network analysis to determine key geological and petrophysical parameters such as lithology, permeability, hydraulic flow units and reservoir quality indicators. These parameters are determined real-time to place wells into the most productive reservoir intervals. This new technique integrates core descriptions, logging and laboratory measured core data from several offset wells to develop cluster based models that are used to train log data to determine rock types and their inherent petrophysical characteristics. The prediction model is then used in a real-time analysis to geosteer wells into high-pay intervals. |
