簡介：IBM和殼牌最近宣布將進行一項聯合研究項目，利用先進的油藏產量預測技術，協調地質與油藏工程方面的現場數據，其目標是獲取剩余油藏得快照，提高油氣采收率。
IBM and Shell recently announced a new joint research project aimed at extending the productive life of oil and gas fields using advanced predictive techniques to reconcile geophysical and reservoir engineering field data. The goal: a more accurate snapshot of remaining reserves that can be extracted more efficiently.

The collaboration will unite IBM's predictive analytics, mathematics, and supercomputing experience with Shell's subsurface and reservoir expertise to create a more efficient and accurate understanding of mature reservoirs and ultimately, an optimized recovery rate for oil and gas.

This is one example of IBM's initiative to reinvigorate large and complex infrastructures using analytics with predictive insights, according to Ulisses Mello, research scientist and manager of IBM's Petroleum Energy Analytics group. "In this particular case with Shell, we are trying to help them optimize their reservoir models that predict future reservoir behavior based on pressure matching from well data.

"This automatic pressure matching to predict reservoir performance has become increasingly complicated as Shell has moved to time-lapse [4D] seismic techniques," Mello continued. "Today's 4D seismic techniques allow snapshots of the reservoir to be taken more frequently, which provides a great deal of data to be incorporated into the model."

Currently, the process of reconciling sometimes-conflicting data harvested from instruments and acoustic measurements takes one month or longer, and involves a cross-disciplinary team: geophysicists must examine time-lapse seismic data from subsurface rock formations; reservoir engineers receive well and laboratory data; and geophysicists receive information in the form of sound waves that cover wide spaces between the wells.

The Shell/IBM project aims to reformulate and automate the task of reconciling the different data sets and create a more elegant mathematical optimization solution. As a result, the reconciliation processing time will be reduced from months to days, according to IBM.

"In addition, rather than doing the matching for the well data alone, we are going to try to do the matching for the seismic data at the same time," Mello added. "That will be very interesting, because then we will be able to fill in more of the spatial resolution compared to well data that is relatively sparse nowadays."

The joint project has an initial span of 2 years, and will involve IBM and Shell research scientists in several laboratories in the US and The Netherlands.

The solution will be trialed on a few Shell assets prior to rolling out on a worldwide basis. Once available, it will become part of Shell’s proprietary reservoir modeling tool kits for application in new oil and natural gas developments as well as existing assets.

For more information about Shell, please visit www.shell.com . For additional background about IBM Research initiatives, please go to www.ibm.com/research .