VOM2MPS: from virtual outcrop models to multipoint statistics training images for improved reservoir modelling

The development and management of oil and gas reservoirs is based on accurate and realistic 3D computer models of the subsurface geology. The geometry of the rock layers and geological features are mapped with limited resolution using methods such as seismic surveys and core analysis, and then assigned physical characteristics relating to how fluid flows through the reservoir. To give more detailed knowledge of geological features, geologists study exposed areas of rock on the earth's surface (termed outcrops), which are representative of the geology in the reservoir. Using state-of-the-art digital mapping techniques, such as laser scanning, to make virtual models of the rock outcrops, geologists can analyse features and ultimately apply new insights back to the subsurface reservoirs. This project aims to use advances in mobile computing to assist geologists in their interpretation of outcrops in the field. Tablets are becoming ubiquitous and are now suitable for visualising and interacting with 3D virtual models. A tablet-based application will be developed that will allow 3D models and new camera images to be interpreted in the field directly. The project will explore how interpretations can be efficiently made in the field, and develop workflows for seamlessly integrating them into existing model data. These interpretations will be used to generate statistical training images for modelling geological features in reservoir models. The project is collaboration between Uni Research, Bergen, and the University of Aberdeen, UK, and will fund two PhD students over a timeframe of 3.5 years. Geology, geomatics and computer science will be married in this interdisciplinary study, which will generate new knowledge on acquiring and using outcrop data for sub surface reservoir modelling. The project forms an extension of the ongoing SAFARI programme, funded by the Research Council of Norway and the FORCE consortium of oil and gas companies operating on the Norwegian shelf.

An immediate outcome of the VOM2MPS project is the training of two PhDs, in geology and computer science, who have published in an international context. New workflows allow generation of training images from virtual outcrops. A novel algorithm has been devised for co-registering 2D and 3D data. This formed part of a mobile application for interpreting field geology, though applicable in other disciplines. This resulted in the receipt of two best paper awards. The project has a tight coupling with the long-running SAFARI programme with its industry stakeholders. Results are being implemented in SAFARI, which will impact operators through new knowledge and data for aiding understanding and development of conceptual models. In a wider societal context, improvements in modelling through multiscale integration of subsurface and outcrop data has the potential to offer more efficient, and therefore greener, use of natural resources, and are transferable to CO2 storage and geothermal energy.

Creation of geocellular models - computer-based representations of the geometry and properties of subsurface geology - is a fundamental task in managing oil and gas reservoirs. Central to this is the stochastic population of models with physical character istics of the rock and dimensions of depositional features. Recent approaches to statistical modelling have favoured multipoint statistics (MPS) as a more realistic way of populating model cells, based on defining training images to express geological con cepts and spatial relationships. A critical issue is the creation of these training images, which may be derived from photographs, sketches or conceptual models. Outcrops are an important but underused source of information that may be harnessed in MPS mo delling. In this study, we propose to link state-of-the-art geospatial mapping methods with field-based interpretation tools for generating MPS training images. An existing database of 3D virtual outcrop models will be used as the basis for geological int erpretation of facies and sedimentary bodies. Advances in mobile computing have the potential to revolutionise field geology. A tablet-based application will be developed, allowing 3D models and co-registered images captured with the device's on-board cam era to be interpreted by geologists directly in the field. The interpretations will serve as the basis for creating a library of MPS training images, and for performing reservoir modelling simulations. Finally, training images and synthetic reservoir mode ls will be stored in the outcrop database, allowing end-users to visualise virtual outcrop models with reservoir modelling products to get an improved understanding of relationships in the outcrop. The project is led by Uni CIPR in Bergen in collaboration with the universities of Aberdeen, Bergen and TU Wien, and is supported by the FORCE consortium of oil companies. The project will involve two PhD students, several master's students, and will run over 3.5 years.