Recently the emphasis on Big Data for oil and gas exploration has been on analytics and data mining from diverse geospatial datasets, as companies with an ageing technical workforce and increasing dependence on technology, attempt to learn in the right place and extend globally to newly accessible areas, using large volumes of acquired domestic data
The oil and gas industry has always been known for being data and technology driven, and by the early 1990’s, exploration for domestic hydrocarbon reserve capacity in Malaysia was already generating petabytes of digital data from geological and geophysical surveys in onshore and offshore oil provinces. In a way it could be argued that oil and gas was doing “Big Data” before it became a buzz word in data management. Recently the emphasis on Big Data for oil and gas exploration has been on analytics and data mining from diverse geospatial datasets, as companies with an ageing technical workforce and increasing dependence on technology, attempt to learn in the right place and extend globally to newly accessible areas, using large volumes of acquired domestic data. The drive to maximize the domestic energy resource capacity of Malaysia and support global growth in export demand has led Malaysian energy companies to look more closely at their large volumes of geospatial data, increase the capacity for Big Data analytics, and find ways to present key metadata in a map view using advanced Geographic Information System tools. This represents a switch in emphasis from just technology, to a full geospatial infrastructure to support mission critical decisions. This paper presents a case study on how advanced geospatial data analytics were used at one oil and gas operator in Malaysia to improve their domestic resource management and derive the maximum benefit from a combination of datasets with large volumes, variety, velocity, and veracity. The business case for using Big Data in a geospatial context was built on a need to analyse the geotechnical data that had supported successful drilling in other provinces, and apply it to projects where Mubadala Petroleum was partnered with PETRONAS in key offshore areas. Applications of geospatial Big Data include analysing, determining, and predicting the effect of rock type, fluid content, and pressure at depth on oil and gas production from offshore fields. The results of the Big Data analytics were made available to geotechnical users through a geospatial interface using a query-based table of contents and a service oriented architecture.
Following industry best practices, Mubadala uses a centralised function for exploration and production technology applications and information management, which enables efficient and directed management of the large data volumes used in the petroleum industry. This function is charged not only with establishing standardized processes for analyzing and delivering data, but also for evaluating new technologies that could enhance the business value of data held as a resource and asset. In Malaysia, PETRONAS as the national oil company retains title to all original domestic data acquired under the terms of a Production Sharing Contract with the operator, but it is in the commercial interests of the operator to find the best way to manage and utilize this data to support mission critical decisions such as where to site wells or facilities.
Almost all data used in the oil industry has a geospatial component, and much of the multiple tens of terabytes of data collected during evaluation of exploration decisions would be virtually useless without accurate geospatial positioning. This is true to the extent that most oil and gas operators combine a GIS function with their geotechnical data management, and many have dedicated geomatic, cartographic and geodetic experts on staff.
Figure 1. Geologic cross section showing the potential effect of a geodetic positioning error on the outcome of an exploration well (From the Oil & Gas Journal).
The geospatial content of petrotechnical data sets has only increased as the volume of data exploded in the 1990’s with the widespread acceptance of 3D seismic data and in the 2000’s with the advent of real-time monitoring and operational information streams. The integrated interpretation of this data in a geospatial context from disparate technical disciplines has set the stage and enabled the business drivers for Big Data in oil and gas.
In addition, geospatial tools allow members of cross-functional asset teams to share knowledge, collaborate on best practices, and capture lessons learned in map views that are both familiar to experienced petrotechnical professionals and accessible to millennial knowledge workers. At Mubadala, a network GIS viewer was deployed that displays query-based layers for three separate classes of data consumers; data technicians, discipline specialists, and managers and function leads. The geospatial content is culled and assimilated from disparate corporate and project data stores, cloud based paid subscription data, and transactional operational feeds, in order to provide timely and validated data support to key business decisions. Distribution over the intranet means that data QC and validation processes can be piloted in high-visibility business units and then adopted and adapted as global standards.
Several facets and dimensions of geospatial oil and gas data make it both a challenge and a technology with high rates of return for data scientists. In addition to the now accepted industry measures of volume, variety, velocity and veracity, oil and gas big data possesses measureable qualities that add to both its complexity and the capability maturity in technology required to manage and extract value from it. These qualities have been previously defined as; propagation (the process by which data volumes are distributed and duplicated by iterative workflows in different disciplines and functional silos within the organization), proliferation (the rapid rate at which data volumes are multiplied by loading and use in specialized tools and applications with contradictory interpretations), pervasiveness (the ability of technical data volumes to expand to fill available storage capacity through the use of multiple working versions, scenarios and realizations) and persistence (the capacity of data add value to oil and gas assets which have generational life spans, even when used to support decadal projects and when supported by budgets with an annual or quarterly relevance). All of these properties can be measured and quantified in metrics collected by geospatial metadata tagging and storage analytics, and used to select and tune the big data technologies that are bested suited to extract business intelligence in a given workflow. In addition much oil and gas data is unique in that it comprises indirect measurements of physical properties with quantifiable but wide ranges of uncertainty, and that corporate geo databases usually contain a combination of proprietary, subscription, shared and public domain data, all with different classifications, formats, and security and retention levels.
Figure 2. Geospatial representation of oil and gas fields, concession block polygons, geological basins and international boundaries in offshore Malaysia waters available from paid subscription services (data courtesy of IHS).
What is clear is that partly because of the huge expense of acquiring proprietary exploration data, and partly because of the high capital cost of the projects that it supports, oil and gas geospatial big data has a much higher visibility as a resource and asset than in other industries. In the case of the huge data volumes acquired domestically in Malaysia and used to support domestic energy capacity and support export growth, much of the value added comes from operator experience in other analogous geospatial settings and operational environments. Practically, this means that big data technologies for geospatial oil and gas data must be based not on a single application or analytic tool as in other industries, but on a purpose-built and enterprise supported spatial data infrastructure that spans disparate disciplines and delivers validated data with known quantitative uncertainty to a decision maker with minimum latency. Because geospatial decisions around exploration and production often involve multiple realizations of data, the spatial infrastructure must be able to support ad-hoc queries from users with widely differing views of the organisation.
In the case of Mubadala’s experience with Malaysian geospatial oil and gas data, large projects were created that allowed synthesis and comparison of domestic data with regional datasets from adjoining geologic provinces in Indonesia, Thailand and Vietnam. Several data types were combined into layers in the corporate GIS system for easier visualization and analysis, and key geospatial metadata was associated with query-based layers and a table of contents that reflected both the technical disciplines that serve as suppliers and consumers of the data, and the Mubadala standardized exploration methodology.
Some of the key displays and analyses that resulted from early application of geospatial big data analysis were diagnostic statistical outliers in a large global dataset containing technical measurements of rock type, fluid content, and pressure at depth in selected wells. While this dataset represented legacy static data, the same methodologies can be applied to real-time measurements from the same data domains as the organisation moves from exploration to production on key Malaysian assets. Much of the analysis is used to support critical decisions about major capital investments with exploration concession blocks held and offered by PETRONAS.
Figure 3. Example of a regional scale geospatial study using field analog and petroleum systems data in a GIS view (courtesy of Schlumberger).
The value of the geospatial big data in these GIS displays is increased by the ability to drill down from key map elements to the business metadata associated with them. This includes not only the variety of technical measurements and reports associated with a well or seismic survey, but information about which subject matter expert validated the data and when, what the legal and commercial terms and restrictions were around the data acquisition and interpretation, and what key decision documents were created from the data to support an internal workflow or stage gate process. In the future it can be anticipated that data types with larger volumes, more variety, faster acquisition rates and more direct relevance to mission critical decisions will be added to these GIS displays.