First and foremost, hats-off to SMi for organizing this, the 17th edition of it E&P data management conference which was held in London earlier this year in the throes of what is probably the deepest industry downturn in the UK to date. Chris Bradley (Independent consultant) reminisced on the days of ‘big money’ data projects when a major could afford $10-15 million on a ‘big bang’ enterprise development. Even then the outcomes were disappointing. Today, there is renewed interest in ‘big data.’ But the problems of data governance and stewardship remain. Most organizations have not sorted out their ‘little’ data. Bradley advocates a Dama-type approach starting with a high level map of the business to identify key concepts that are shared across exploration, production, downstream and retail. This provides a framework for a conceptual data model, itself the basis for more drill down into detailed objects and complex relationships. You also need to understand data ownership, and ‘prepare to be horrified’ as some data may be held in ‘27 different systems.’ Data management is not the ‘field of dreams,’ but rather the ‘art of the possible.’ Plan big, implement small. Data models and governance are vital.
While we were reflecting on the vitality of data governance and ownership, Neil Storkey (BG Group) cited ‘ownership,’ along with a ‘single version of the truth’ as two of his pet hates! ‘We don’t want ownership in our vocabulary.’ Bradley’s Dama approach has failed while Storkey has had ‘pockets of success.’ But pockets are not enough, a new data order is needed. First try to get IT out of the loop and stop switching to the next new technology panacea. What is required is more attention to strategy, Storkey’s specialization. This means adopting the RACI framework and getting the business to say what outcomes are needed. BG’s data LifeSaver defines responsibilities of data providers and consumers. Next you should empower citizen stewards of data, ‘even though Gartner says they will fail!’
Peter Jackson provided a plug for Bain & Co.’s Rapid toolset for clarifying decision accountability and its application to big data projects. According to a Bain study, only 4% of companies ‘execute well’ on big data strategy and oil and gas particularly lags on adoption. This is due to a lack of understanding of big data’s capability and an unwillingness to change how we work. Companies should create a big data center of excellence and staff it up with data scientists, analysts, developers, data managers, engineers, product managers and business/legal folks (blimey!). Alteryx’ workflow-based cleansing and Tableau both got plugs as key analytical tools used inter alia to identify pump seal failure, fluid contamination and out-of-spec operations. All achieved with data mining, regression analysis and fuzzy logic.
Magnus Svensson (Dong E&P) traced Norway’s rather intricate history of production reporting and the EPIM standards body. There is more to production reporting than meets the eye as oil and gas flows from a subsea module, into a production platform and on through export pipelines to the shore. The process involves complex volume reporting by the operator to partners and by partners to the government. There has been a decade or so of various production standards development and ‘we are still not finished.’ Reporting still involves manual massaging of data, but it is getting harder to tamper with data now it is not just in text format. This is an important consideration as a terminal operator may report for 20 different companies. Despite the current standards’ imperfections, implementation has been a worthwhile, albeit costly exercise. The authorities are keeping a close eye on developments as the industry ‘may have been reporting the wrong numbers for 10 years.’ In the Q&A it emerged (to some laughter) that a parallel UK initiative, PPRS, the production data exchange format has ‘lost focus.’
Jill Lewis (Troika) reported on another standard that was having trouble with adoption. SEGY Rev 1 has been out for 13 years and is still hardly used. SEGY rev 2 will be out ‘real soon now.’ Even with a standard, individuals tend to do things differently so companies need to automate checks and leverage expertise. Data management is a vast subject and you need to prioritize. All field data should be in SEGD. But some companies are still using formats from 1975! ‘You can’t be serious!’ Specifying a required format in a contract is a necessary but not sufficient condition. One company specified SEGD Rev3 in the contract. But the data came back in Rev 1 (an old contract had ben used). Format and QC checks (with Troika’s software) are particularly important before old tapes are destroyed. ‘Make sure that you have control of this stuff even though your outsourcing partners may not relish your efficiency.’
Jonathan Jenkins (NDB) and co-author Cindy Wood of the UK Oil & Gas Authority presented on production data management, allocation and reporting. Recent Guidance Notes for Petroleum Measurement heralded an OGA review of production allocation systems. These dovetail with a 2016 Aupec/NDB data maturity survey which found that most companies have no dedicated production data management at all and simply rely on Excel. One result is that engineers don’t trust allocation figures. Volumes can be 30-40% adrift. Deployment at client sites is almost exclusively a choices between different vendor’s software. ‘Standards? They could not care less, so we did not do it.’ Jenkins’ goal is to ‘get production data out of the closet,’ with an objective of saving 50 days per engineer per year by building a central body of truth exposing the same figures to both allocation and production systems.
Tor Jakob Ramsoy (Arundo Analytics) sees a ‘perfect storm’ about to hit oil and gas, driven by the internet of things, the cloud and by data science. At the nexus of the storm is Arundo’s offering. Companies need to ready themselves for the storm by high grading their CIO and adopting a digital strategy. Deep learning is exploding across the whole industry. Data, not software is the real challenge. Ramsoy advocates data transparency across value chain. FMC, Aker and Schlumberger can all provide embedded software and an integration layer to support a cloud-based data mart. Asset owners will subscribe to apps running against the data platform.
Robert Bond (Charles Russell Speechlys) provided a salutary reminder that data, big or small, is subject to a host of regulations covering its protection, cyber security and due diligence requirements in M&A. These requirements differ from country to country. If big data is where the money is, folks will try to steal it! So rule number one is, if you don’t need it, get rid of it! Other issues arise on the transfer of data from one company to another. EU legislation requires that the recipient of such a transfer must be able to demonstrate its right to use the data for specific purposes. In a merger, the US is particularly watchful about divulging personal data to non-US entities. Some legislations have extremely prescriptive laws including the possibility of sending you to jail. Some companies like PeopleSoft and Salesforce used to rely on US safe harbor legislation for data in US the US. But safe harbor ended in October 2015. Roll in social media, home working and BYOD* and you are entering a minefield!
Dumitru Roman presented Sintef’s DataGraft cloud-based service that offers developers ‘simplified and cost-effective solutions for managing their data.’ The system went live late last year and allows users to transform tabular data into semantic-web style RDF**. Roman opined that even ‘open’ data such as that provided by the NPD Fact Pages is hard to query and hard to integrate with other Norwegian datasets such as the Norway business registry or publically available real estate data. Roman’s answer is to ‘scrape’ data from the PDF sources and create a cloud-based data service. DataGraft is a part of the linked open data movement and assumes familiarity with W3C standards like RDF, Sparql and Owl. Data can be hosted on DataGraft’s semantic graph database. The work stems from the EU 7th Framework DaPaaS project.
ArcaStream’s James Pitts enthusiastically announced that he was going to talk about technology. Specifically, about ArcaStream’s software defined storage and leverage commodity IT pricing. Often, vendor data management strategies are over-reliant on human intervention and you end up with a ‘data junkyard.’ Tiering, replication and other data services may only work with the vendor’s solution, leading to lock-in. Other ArcaStream tools automate data ingestion, extracting and indexing metadata from a magnetic the tape. (Some seismic data specialists appeared skeptical of this possibility when confronted with real world ‘warts-and-all’ seismic tapes!)
Pat Schiele (GE)
thinks that, in the downturn, the upstream can learn a lot from the
downstream especially in using data for asset maintenance. The
downstream, along with power gen and aviation have all ‘been here
before’ and have moved from time-based to condition-based maintenance.
In the upstream though, every well Christmas tree is custom, making for
high cost of design, manufacturing and maintenance. Schiele argues that
more standard designs should stay within a ‘range of suitable’ as
opposed to the current ‘every well is different’ approach. He also
observed that the current practice of paying a company like GE for
maintenance means that the ‘cost/reward structure is wrong.’ We need
more data sharing and collaboration. This happens in utilities and
power gen, not so much in the upstream. ‘We can make money at $40 oil, just not very much.’
* bring your own device.
** resource description format.
© Oil IT Journal - all rights reserved.