Reservoir properties prediction in carbonate reservoirs

• Main Author: AlBinHassan, Nasher M.
• Other Authors: Wang, Yanghua
• Published: Imperial College London 2010
• Subjects: 550
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.520357

Prediction of fluid pressure before drilling, using geophysical methods helps the industry a lot in saving human life, drilling hazards, and equipments.There are several geophysical methods available to predict the fluid pressure before drilling but the most commonly used in the industry are those based on seismic velocities. However, seismic velocities methods are applied on clastic reservoirs with the assumptions that the pressure mechanism is due to mechanical compaction. A major exploration challenge is to successfully predict the presence of high pressure zones in the carbonate reservoirs. Carbonate reservoirs have a more complicated internal structure than clastic reservoirs. The main objective of this study is to predict the carbonate reservoir properties such as porosity and fluid pressure. The new prediction methods that I used in this thesis are called the artificail intelligent algorithms. These algorithms are better than the conventional geophysical methods because of their ability to explore complex relationships between the input seismic attributes and the predicted properties. The algorithms include artificial neural networks and group methods of data handling. Empirical equations from seismic prediction methods were used to transform velocities to fluid pressure. High resolution velocites (wavefrom tomography) proved that better prediction can be achieved when using better input velocity. The velocity methods performed a nice prediction when used with clastic seismic data but proved to give unreliable results when used with the carbonate seismic data. This was because of the difficult internal structure of carbonate reservoirs. The neural network methods proved that they are robust in clustering and segmenting the input carbonate seismic data. The usage of more input seismic attributes made the neural network methods better than the conventional velocity methods. Also, this gave the neural network methods more information about the same physical reservoir property. Among the different seismic attributes used in the experiment, seismic inversion and coherence attributes showed good reaction to high pressure zones. Porosity results from the supervised neural network method were used as a guide to the unsupervised neural network method to predict fluid pressure. The group method of data handling algorithm is performed here for the first time with seismic data to predict the reservoir properties. The new method showed faster and easier prediction than the neural network methods. The automation of the new method yields to better porosity and pore pressure prediction.