关键词: 克拉玛依油田, 冲积扇, 训练图像, 地质矢量信息, 多点地质统计学

Abstract:

 As alluvial fan reservoir is typical in non-stationary distribution, the traditional modeling algorithm based on the steady-state assumption cannot simulate it accurately. On the basis of full analysis of the reservoir distribution, a case study of alluvial fan reservoir on the Triassic Lower Karamay Formation, Yizhong Area, Karamay Oilfield was carried out to integrate modern sediments, underground dense well group, outcrops and other information, using the latest Vector Information based Multiple-Point Simulation to reproduce a fine and reasonable three-dimensional geological model of reservoir architecture, which can be used as a reference for similar reservoir stochastic simulations. First, based on modern sediment and underground alluvial fan reservoir architecture analysis, an alluvial fan training image in line with the study area was constructed under the instruction of pattern guiding and scale constraints. Second, the vector coordinate system of an alluvial fan was constructed. Then the vector information of both training image and the target modeling area was characterized in this system. Finally, the reservoir architecture was reproduced by using the Vector information based on the Multiple-Point Simulation method. The reliability analysis of the realizations was performed. The result shows that the realizations are faithful to well data. Alluvial fan facies distribution in the model is suitable to geological model. Architecture element characteristics including shape configuration, distribution, combination style and scales matches training image and understanding reservoir architecture acknowledgment perfectly.

Key words: Karanay Oilfield, alluvial fan, training image, geological vector information, non-stationary, multiple-point geostatistics