当前位置: X-MOL 学术SIAM/ASA J. Uncertain. Quantif. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
A Probabilistic Clustering Approach for Identifying Primary Subnetworks of Discrete Fracture Networks with Quantified Uncertainty
SIAM/ASA Journal on Uncertainty Quantification ( IF 2.1 ) Pub Date : 2020-04-22 , DOI: 10.1137/19m1279265
Dave Osthus , Jeffrey D. Hyman , Satish Karra , Nishant Panda , Gowri Srinivasan

SIAM/ASA Journal on Uncertainty Quantification, Volume 8, Issue 2, Page 573-600, January 2020.
Fractures form the main pathways for flow of fluids and transport of constituents carried by these fluids in fractured subsurface media. The majority of flow and transport occurs in the primary subnetwork or backbone, which is a subset of the fracture network. Understanding characteristics of the fracture network flow and transport backbones is thus of great importance towards improving efficiency of subsurface applications, such as aquifer management, hydrocarbon extraction, and long- term storage of spent nuclear fuel. We propose a method for identifying these backbones that also quantifies uncertainty over all possible backbones. Our approach treats the backbone identification problem as a probabilistic clustering problem. We develop a probabilistic generative model to simulate backbones that are induced acyclic flow networks. A user-controlled parameter determines the size of the identified backbones, with small backbones having good agreement with early particle arrival times and larger backbones having good agreement with early and late particle arrival times. We demonstrate our method on a representative fracture network with 264 fractures. Our method discovers flow channelization in backbones roughly 10%--50% the size of the original network, supporting previous experiments and discrete fracture network simulations. To the best of our knowledge, our method is the first backbone identification method to quantify uncertainty over backbones.


中文翻译:

概率不确定性离散骨折网络主子网的概率聚类方法

SIAM / ASA不确定性量化期刊,第8卷,第2期,第573-600页,2020年1月。
裂缝形成了流体流动和这些流体在破裂的地下介质中输送成分的主要途径。大部分流量和传输都发生在主要子网或主干网中,而主子网或主干网是裂缝网络的子集。因此,了解裂缝网络流动和输送主干的特征对于提高地下应用的效率(例如含水层管理,碳氢化合物提取和核废燃料的长期存储)非常重要。我们提出了一种识别这些主干的方法,该方法还可以量化所有可能的主干的不确定性。我们的方法将骨干识别问题视为概率聚类问题。我们开发了一个概率生成模型来模拟诱导非循环流动网络的骨干。用户控制的参数确定所识别骨架的大小,小的骨架与早期粒子到达时间有很好的一致性,而较大的骨架与粒子早期和晚期到达时间有很好的一致性。我们在具有264个骨折的代表性裂缝网络上展示了我们的方法。我们的方法发现骨干网中的流通道化约为原始网络大小的10%-50%,从而支持先前的实验和离散裂缝网络模拟。据我们所知,我们的方法是第一种量化主干不确定性的主干识别方法。我们在具有264个骨折的代表性裂缝网络上展示了我们的方法。我们的方法发现骨干网中的流通道化约为原始网络大小的10%-50%,从而支持先前的实验和离散裂缝网络模拟。据我们所知,我们的方法是第一种量化主干不确定性的主干识别方法。我们在具有264个骨折的代表性裂缝网络上展示了我们的方法。我们的方法发现骨干网中的流通道化约为原始网络大小的10%-50%,从而支持先前的实验和离散裂缝网络模拟。据我们所知,我们的方法是第一种量化主干不确定性的主干识别方法。
更新日期:2020-04-22
down
wechat
bug