Shell Exploration and Production has global teams working on new business, exploration and field development projects. A review in the business during 2003 found a significant amount of staff time was spent looking for information rather than doing technical work. For hydrocarbon field development projects, there were also opportunities for improvement around the handover process from one team to another. When projects were put on hold and then later restarted forming a new team, finding all the relevant definitive documents and data was problematic.

Business requirements
The business wanted a global, not point- or region-specific, solution. New business, exploration and field development wanted the same approach so the baton could be passed seamlessly from one team to another. Any new system had to include data, documents (electronic, hardcopy) and processes, and internal and external content. In its simplest form it had to be intuitive to use without training.

Information management landscape
A number of global Information Management (IM) improvement projects were initiated to

Figure 1. Shown is interface used for searching. Connection to map systems also allows spatial searching. (All images courtesy of Flare Solutions Limited)
address these business issues (and others). Building a skilled pool of competent IM professionals internally was critical as well as providing more intimate services to help business teams. Targeted, large-scale data quality improvement initiatives were combined with document scanning projects to digitize selected legacy hardcopy material going back 50 years.

For high-precision, structured search using business metadata, the Web-based exploration and production (E&P) Catalog from Flare Solutions was used. The software was also deployed for high-precision publishing and tracking of work products, integrating with the Microsoft desktop and OpenText Livelink Electronic Document Management System (EDMS).

The software provides a metadata layer and end-user graphical user interface to connect and complement a company’s existing information sources. This complements the full text search engine used by the business, which gives high recall of information but does not always enable staff to easily locate the specific information they need in their project or domain context. The software was also integrated with existing data query and map systems — Geographical Information Systems (GIS) — so content could be found spatially.

Cataloguing philosophy
The philosophy is “high-grading information that needs to be kept for the long term.” Managing the key deliverables, snapshots and supporting material at every step of the business process is the challenge. Care was taken not to focus on quick wins. Uniquely, all content including data (views of data in underlying relational databases), digital application archives, physical items, documents, well files, maps, processes, even people were catalogued. Each was treated in the same way, using the same metadata and attributes for a holistic, high-precision view of what exists.

Content is published into the metadata catalogue through two main routes. First, E&P technical and support staff add keywords from structured lists to information items. Second, content is automatically and manually uploaded (using Web services via a Services Oriented Architecture [SOA] approach) from internal repositories and archives. This includes external information that the business is entitled to and metadata on what is available to buy.

E&P taxonomy
The backbone of the approach is the E&P taxonomy used by the software, which contains hundreds of thousands of technical and business terms, aliases and relationships. This allows manual and automatic tagging of information to aid information retrieval and understanding of the overall E&P business. Taking a simple geological example, if the system finds the term “glauconite” in the original metadata (perhaps a legacy library index), it understands this is a mica, a mineral, discipline mineralogy and usually indicative of marine continental shelf depositional environments. Automatically adding these types of relationships enables staff to discover relevant information that they otherwise would miss.

High-precision structured search

Staff can search for information via four main methods (or combinations) to slice and dice the corporate memory (Figure 1). The first three focus on “I know what I want” searches; the fourth is for a serendipitous “I am not sure what I want” browsing approach.
1. Type in a term and press . This gives back a high-precision list of what exists within the business and from external sources. This mode is similar to a typical search engine.
2. Pick terms off structured pick lists such as well, field name, project name, product type (e.g., well proposal, pressure data). Pick lists contextually filter as you select items, narrowing down choices automatically. These can be combined with the bibliographic metadata (author, owner, published date, etc.) similar to a library way of searching. This is particularly useful if you know exactly what you want: “Show me VAR2 outputs for the Amethyst project” or “Show me Field Development Plans (FDP) submitted for Amber Cluster Development over the past 4 years” or “After Action Reviews (AAR) for Nigerian projects at Phase III.”
3. Personalize saved searches with dynamic folder structures where the same item can appear in different folders. Normal folder structures only allow a single hierarchical representation to organize information — it’s impossible to satisfy everyone! Saved searches can be particularly useful for well files, where many disciplines want their view on the information. This is similar to Windows Explorer navigation.
4. Graphical search results are distributed across the E&P taxonomy (BrightSky). The user can interactively drill down, clicking on a cell at any time to perform a high-precision search. For the first time, this enables staff to graphically navigate results lists to see what exists against an E&P taxonomy. This includes internal content and external information.

High-precision publishing
One of the key elements of publishing is via a process template. The product contains the best practice for the Opportunity Realization Process in the business. This contains the steps for field development, with explicit steps for key deliverables ensuring a business audit trail with approvals from the various boards for compliance.

The system acts like a publishing assistant, containing pre-defined keywords from the E&P
Figure 2. Process structure can be seen. By clicking on the appropriate step (containing predefined attributes), staff can publish documents, which are automatically put into safe areas of the company’s document system.
taxonomy reducing the burden on the technical professional and manager. The organization of information from a project perspective allows the team to easily find definitive documents and supporting information while also allowing information to be found through general searching. Interactive, graphical-view dashboards allow drill-down by date, status, process and project (Figure 2).

Results
Since going live in Shell in mid-2005, more than 4,000 staff members are using the system in four continents. The catalogue contains more than 1.8 million items today. An independent, end-user survey during 2006 indicated it had contributed to a 50% reduction in the time spent by staff looking for information. This equates to tens of millions of dollars in efficiency savings. In addition, new information discovery using the system has directly led to the identification of new plays and potential revenue generation. Global rollout to the top 70 field development projects in Shell is still ongoing. However, early signs indicate significant improvements are being made.