2013年10月18日-19日，参加了西安第三届数字油田高端论坛暨第二届国际学术会议，本次会议主题是大数据和数字油田2.0，参加本次会议的目的是想看看大数据这个概念如何在油田领域中落地，但很可惜与大数据有关的报告或者与石油没有关系，或者只是标题中有个大数据，但内容还是关于数字油田建设进展。只有程国建的一篇介绍SPE数字能源大会的材料中讲到了一篇国外报告，可能介绍了大数据在油田可能的应用领域，回来再仔细读读这篇材料。不管如何，把几个感兴趣的材料的主要思路整理了一下。

王璞（谷歌）：大数据处理在google----全球数据分析的方法技术

介绍了谷歌大数据处理架构中的几个主要概念：MapReduce，Shuffle，BigTable/GFS，XFE（extensible front end），protocol buffers & stubby（通讯协议，可将protocol buffer转换为程序代码）。

高灯亮（美国西佛尼亚大学）：地震属性在油藏描述中的三维可视化技术

报告比较清晰，主要就讲了三部分内容：Reservoir structures构造，Reservoir facies相，Reservoir properties属性，对于他的属性分析的算法不了解，但他把不同的相facies用不同的颜色和透明度来三维表示，一些三维可视化的效果还是值得我们学习的。

石玉江（长庆油田）：大数据与油藏数字化----油气藏数字化协同研究和决策支持平台的建设与展望

是长庆油田一个6年项目的进展报告，不评论，提到的几个技术：业务流程标准化，SOA框架，服务总线，数据链技术，数据整合，专业软件接口，地质图件导航，远程传输，协同网络化研究环境。

张志檩：从两化融合看数字石油石化的技术内涵

报告中有很多国外的素材，可惜时间有限每张片子都过得飞快，报告也不让拷贝。

张海：大数据处理的统计方法----油田大数据分析方法

介绍了大数据的一些概念，举了2个例子，一个是社交网络的，一个是与猴子有关的，很遗憾都与油田没有关系。

张文坡：辽河油田的油田大数据建设与展望

讲的内容主要是六个“大”，大数据中心（数据建设）、大计算（云桌面）、大存储（盘阵）、大网络、大系统（数据库集群）、大应用（油气水井生产数据管理、生产调度指挥、物联网、ERP、企业门户）。

程国建：数字油田国际动态----2013 SPE数字能源大会专题介绍

20分钟概览了SPE大会的主要内容，我主要感兴趣的：

SPE-163718，Digital Oil Field Experience: An Overview and a Case Study

斯伦贝谢的一个数字油田案例实施的经验和教训

SPE-163709，Design of an Automated Drilling Prediction System - Strengthening While-Drilling Decision Making 自动钻井预测系统的设计----强化随钻决策

SPE-163683，The Roadmap for Industry Adoption of the PRODML Standard

PRODML标准的工业部署路线图

SPE-163717，大数据大买卖 Examples of Work with Big Data

Work on the application of Big Data and analytics in the oil and gas industry is in the experimental stage. Much of the work centers on data-intensive computing and how I/O data loading can be managed most efficiently. Such as:

• Use of Hadoop to process seismic data. Chevron is using Hadoop as one of the 25 steps in the workflow for the identification of reservoirs. Processed data is fed into high-performance computing models. The project uses the IBM BigInsights technology, which includes the Hadoop component stack.

• Use of Hadoop in the cloud. Royal Dutch Shell is piloting Hadoop in a private Amazon cloud.

• Production data for performance forecasting. One oil and gas company is experimenting with the time series analysis of production data. Aging wells where the forecast does not meet a predetermined production threshold are flagged for immediate remediation.

• Investigation of two MapReduce approaches applied to drilling data. In this experiment, Chukwa, an open source data collection system built on top of Hadoop, was found to be a preferable approach to a Hadoop distributed file system when working with large files.

• Storing and processing seismic data in a Hadoop cluster. Cloudera, a company that provides a data platform built on top of Apache Hadoop, has initiated a project called Seismic Hadoop project.

• Seismic and drilling data in the cloud. PointCross Inc. has introduced two cloud-based offerings for the oil and gas industry — a seismic server and data repository that uses NoSQL and Hadoop technologies to store and manage SEG Y files and a drilling data server and data repository that can accept WITSML, LAS, and WITS formats.

大数据在油气领域的可能应用点Possible uses of Big Data and analytics in the oil and gas industry

• Exploration -- By applying advanced analytics, such as 模式识别pattern recognition, to a more comprehensive set of data collected during seismic acquisition, geologists may be able to identify potentially productive seismic trace signatures that have been overlooked.

• Development -- Big Data and analytics could aid oil and gas companies in acreage assessment and prospect generation. Analytics applied to geospatial data, news feeds, oil and gas reports, or other syndicated feeds could provide competitive intelligence on where to submit bids for leases.

• Drilling -- Beyond monitoring and alerting based on limited data, Big Data and analytics could be applied to real-time "big" drilling data to identify anomalies based on multiple conditions or predict the likelihood of drilling success.

• Production Operations -- Enhancing oil recovery from existing wells is an objective of many oil and gas companies. Analytics applied to a variety of Big Data at once — seismic, drilling, and production data — could help reservoir engineers map changes in the reservoir over time and provide decision support to production engineers for making changes in lifting methods. This type of approach could also be used to guide fracking in shale gas plays.

• Maintenance -- Predictive maintenance is not a new concept for the oil and gas industry, although if you ask a maintenance executive, it does not get the attention and budget it deserves. In upstream, if pressure, volume, and temperature can be collected and analyzed together and compared with the past history of equipment failure on a compressor, for example, then alerts can be automated. The same type of situation is found in midstream pipelines. This would be useful in cases where time to condition detection to failure is short and where assets are considered critical to the operation or failure would have a significant impact on health, safety, and environment.

刘志忠（大港油田）：深化数字油田建设，推进数字油田发展

有些比较实际的认识，不一味地追求技术的先进性，把一些技术概念要真正落地。数据标准化与持续投入才能保证数字油田的持续发展。

甘腊梅：延长油田数字油田建设与数字化协同工作平台

里面有张图竟然用了我们在2006年7月照的一张勘探项目部会议室的照片，里面是于总等人召开生产讨论会时的背影。

再后面的报告主要是一些大会赞助商的产品与广告。

本文转自申龙斌的程序人生
博客园博文，原文链接：http://www.cnblogs.com/speeding/p/3379960.html，如需转载请自行联系原作者

http://www.cnblogs.com/speeding/