多属性神经网络地震反演在NB油田水平井钻探中的应用

An application of neural-network seismic inversion with multiple attributes to horizontal drilling in NB oilfield

NB油田位于渤海海域,为新近系河流相稠油油田,储层横向分布不稳定,砂体厚度薄、连通性较差,油水关系复杂,开发设计部署了大量水平井。选择神经网络反演方法,建立地震属性与储层参数的非线性关系,进而根据其关系反演得到储层参数数据体,利用该数据体沿层切片提取储层预测信息,研究储层发育规律,指导该油田水平井的部署和钻探,取得了较好的效果,降低了钻探风险。用神经网络地震反演方法进行储层预测研究,指导水平井钻探,在渤海海域尚属首次。

NB oilfield located in Bohai Sea is a heavy field in Neogene fluvial facies, and its reservoir is characterized by instable lateral distribution, thin sandbody, relatively poor connectivity and complicated oil-water relations, resulting in a development plan to drill a large number of horizontal wells. The method of neural-network inversion is selected to establish a tween seismic attributes nonlinear relationship beand reservoir parameters,by which a data volume of reservoir parameters can be obtained. Then this data volume can be used to extract information for reservoir prediction along its layer slices and to research reservoir distribution. The results have been applied to guide planning of horizontal wells in the oilfield, making the drilling risk reduced. It is the first time to apply the neural-network seismic inversion to reservoir prediction and drilling guidance of horizontal wells in Bohai Sea.