A Methodology for Effective Earth Model Management while Geosteering

The aim of geosteering is to place a well optimally in the reservoir, based on the available measurements and interpretations of the geology. The interpretations are captured in an earth model that is used for decision support. However, real-time decision processes such as geosteering are poorly supported by current earth modelling methods. The geological structure is challenging to modify in a timely manner, and complex structural and stratigraphic uncertainties are typically not well accounted for. When drilling in complex geological settings, such limitations may lead to poor decisions. Current 3D modelling technologies are typically based on an inflexible, globally defined corner-point grid used to represent geological structures and petrophysical properties. The management of this grid is computationally expensive. To pave the way for a more effective modelling methodology, we propose novel principles for i) local updates of the geological structure in an existing grid, ii) local control of the resolution of the geological structure and consequently the grid, and iii) local scale uncertainty management including the geological structure. The principles are demonstrated in a 2D software prototype. The aim is to enable an always updated multi-realization 3D model at optimal resolution while drilling, suitable for real-time decision support under uncertainty.