• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>2001: the energy odyssey >Estimating Reservoir Permeability in Three Dimensional Space by a Versatile Artificial Neural Network Model
【24h】

Estimating Reservoir Permeability in Three Dimensional Space by a Versatile Artificial Neural Network Model

机译：用通用人工神经网络模型估算三维空间中的储层渗透率。

获取原文
获取原文并翻译 | 示例
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Estimating of reservoir properties is a criticalrnelement of resrvoir management and development. One ofrnsuch properties-permeability-is probably among the mostrnproblematic ones to estimate. Generally, prediction ofrnpermeability is achieved by utilizing well tests or byrnutilizing the relation between well log data and corernanalysis data, either by emprical formulas or byrnstatistical techniques. A newly developing alternative tornuse of formulas or statistical techniques is the use ofrnartificial neural networks (ANNs). In this alternative, thernANN is fed with the log data to match it to core analysisrnresults. Yet, such ANN estimation generally has therndisadvantage of requiring too much data and fine-tuningrnby ANN experts for each and every application and field.rnToo overcome the above problems and to furtherrninvestigate the possibility of using ANNs as tools forrnestimating reservoir-wide permeability in threrndimensional space at locations even without actualrndrilled wells, this study suggests the use of a novelrnmethod instead of conventional approach ANNs. The socalledrnneighborhood approach is a new way of usingrnANNs to estimate permeability by emphasizing thernimportance of neighbor points in a more aggressivernmanner. The study demonstrates that such informationrncan improve the quality of the permeability estimationsrnobtained on well-based predictions when compared tornconventional approach ANNs. The study also suggests arntechnique to estiamte reservoir-wide permeability for 3Drnestimation for portions of the reservoir without actual
机译：储层性质的估算是储层管理和开发的关键。此类属性之一（渗透率）可能是最容易估计的问题之一。通常，渗透率的预测是通过利用测井测试或通过利用经验公式或统计学方法来利用测井数据与岩心分析数据之间的关系来实现的。新兴的替代公式或统计技术的替代方法是使用人工神经网络（ANN）。在这种替代方案中，将日志数据馈入rnANN，以使其与核心分析结果相匹配。然而，这样的ANN估算通常具有缺点，即每个应用和每个领域都需要太多数据和需要ANN专家进行微调的缺点。即使在没有实际钻井的地方，该研究也建议使用新颖的方法代替常规方法的人工神经网络。所谓的邻域方法是一种通过使用ANN来通过以更具侵略性的方式强调相邻点的重要性来估计渗透率的新方法。研究表明，与常规方法的人工神经网络相比，此类信息可以提高基于良好预测的渗透率估计的质量。这项研究还提出了对整个储层渗透率进行估算的技术，以便对3个储层进行局部模拟。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号