Study On Describing Reservoir Spatial Properties Variation By Integration Of Well And Seismic Data

In the oil/gas field exploration and exploitation phase, the target of reservoir study is to build a quantitative 3D geological model of the reservoir, which is not only the need of development of oil/gas exploration and exploitation, but also is the need for improving oil recovery. Most of the fields in China are in the middle-late development or a mature phase. This makes the production of remaining hydrocarbon more and more difficult to develop the remaining oil/gas. It is a major technical issue on how to use geological, log, seismic and other information which have different resolutions, different scales of space covering, to build a precise reservoir model in order to develop the remaining oil/gas and increase enhanced oil recovery. Investigations in the industry have continuously developed various techniques to reduce the uncertainty of the 3D reservoir model, are engaged in unremitting efforts and have done a lot of work.In this thesis, careful study of several commonly used methods of modeling and the comparison of different methods have been conducted. I in order to make use of the capability of seismic data and vertical high-resolutiosn of log, we investigate a new method: seismic derived modeling method. This method is different from the current commonly used methods, and is deterministic approach. First, when seismic data and log data are the same types, for example, amplitude, we used this method to enhance seismic resolution with well logs. Second, seismic data and log's have different data types, we use this method to build a model of properties from log by using support vector machines (SVM). Finally,its result has been compared with Kriging, Co-Kriging, Collocated- Kriging and Sequential Gaussian Simulation and Sequential Gaussian-Bayesian Sim- ulation.In the thesis, synthetic and field data have been used to validate the method propped.