石油与天然气地质 ›› 2023, Vol. 44 ›› Issue (1): 238-246.doi: 10.11743/ogg20230120

• 方法技术 • 上一篇    

基于相序的多点地质统计学相建模方法及其应用

王鸣川(), 段太忠   

  1. 中国石化 石油勘探开发研究院,北京 102206
  • 收稿日期:2022-05-30 修回日期:2022-09-02 出版日期:2023-02-01 发布日期:2023-01-13
  • 第一作者简介:王鸣川(1985—),男,副研究员,储层建模与油藏数值模拟。E-mail: wangmc.syky@sinopec.com
  • 基金项目:
    国家自然科学基金青年科学基金项目(41702359);中国科学院A类战略性先导科技专项(XDA14010204);国家科技重大专项(2016ZX05033-003-002)

Facies sequence-based MPS reservoir facies modeling algorithm and its application

Mingchuan WANG(), Taizhong DUAN   

  1. Petroleum Exploitation and Production Research Institute,SINOPEC Beijing,102206,China
  • Received:2022-05-30 Revised:2022-09-02 Online:2023-02-01 Published:2023-01-13

摘要:

目前的多点地质统计建模方法均以一定维度的数据模板为载体,将概率或几何距离作为建模工区未知区域模拟取值的判定标准,该判定标准在地质建模时与地质含义的联系较为薄弱。针对上述不足,文章回归地质建模的本质,考虑相建模过程中数据事件与训练图型对比的地质含义,提出基于相序的多点地质统计相建模方法。该方法以具有沉积学意义的相序为基本的相似度对比单元,新建适合于相序相似度对比的动态数据模板,采用句法模式识别技术,通过动态规划方法计算数据事件中已知相序与训练图型中对应相序的相似度,对建模工区未知区域进行相模拟,实现了从沉积学意义出发对强非均质性储层进行多点地质统计学建模。理想模型和塔河X区三角洲前缘储层沉积相模拟表明,本文方法能够有效重现训练图像所显示的各地质体的几何形态,较好地反映储层不同相的沉积展布规律,大幅提高了相建模的精度。该方法将沉积学与多点地质统计学有效结合,提出了一种全新的多点地质统计学相似度对比方法和多点地质统计学建模框架,为复杂储层的相建模提供了一种新的方法。

关键词: 相建模, 建模算法, 相似度对比, 相序, 数据模板, 多点地质统计学, 塔河油田

Abstract:

The current multiple point geostatistical (MPS) modeling methods take the probability or geometric distance, which is inferred from training image based on a certain dimension of data template, as the criterion for determining the simulated value in the unknown area of the simulation grid. The relation between the criterion and geological meaning is weak in geological modeling. In view of the shortcomings of geological connection in the previous MPS modeling methods, returning to the essence of geological modeling, considering the sedimentological significance and geological implication of the similarity comparison between data events and training patterns during the facies modeling process, a facies sequence-based MPS modeling method is proposed. The method takes the facies sequence with sedimentological significance as the basic similarity comparison unit, creates a new variable data template suitable for facies sequence similarity comparison, utilizes the semantic recognition technology, directly calculates the similarity between the known facies sequence in the data event and the corresponding facies sequence in the training pattern, and then simulates the facies of unknown area in the simulation grid, through which realizes MPS modeling of strongly heterogeneous reservoirs from the sedimentological significance. The sedimentary facies simulation results of ideal models and the delta front reservoir in Tahe X area show that the proposed method can effectively reproduce the geometric shape displayed in the training image and well reflect the deposition and distribution of different facies in the reservoir, and greatly improve the accuracy of facies modeling. The method effectively combines sedimentology with MPS, and puts forward a novel similarity comparison means and modeling framework of MPS, which provides a new method for facies modeling of complex reservoirs.

Key words: facies modeling, modeling algorithm, similarity comparison, facies sequence, data template, multiple point geostatistics (MPS), Tahe oilfield

中图分类号: