GSA Annual Meeting in Phoenix, Arizona, USA - 2019

Paper No. 202-14
Presentation Time: 9:00 AM-6:30 PM


AUSTERMANN, Gregor, KLING, Melanie, EMONDT, Pascale Dominique and HILDENBRAND, Anne, Institut für Geowissenschaften, Ruprecht-Karls-Universität Heidelberg, Im Neuenheimer Feld 234, Heidelberg, 69120, Germany

Detailed knowledge of clastic succession properties, such as TOC, thickness, porosity, composition or geomechanical behaviour, allows for better results in calculation and interpretation of basin models. At the field scale, these properties control, for instance, hydrocarbon migration, connectivity or fracture aperture and length. This presentation introduces an open-source big data tool, specifically designed to upscale very large (50,000+) well datasets for reservoir and basin modeling purposes. This new tool works in four steps: (a) identification of property boundaries and break down of any wireline log into a facies column, (b) automated calculation of facies thicknesses and subsequent quantification of important descriptors and factors, including thickness histograms, ratios and noise, (c) a user guided quality control (UQC) that refines and enhances the results of the automatic recognition of beds and noise, and (d) an export that can be easily implemented into Petrel, GoCAD, ArcGIS, Petra, Geoscout and other reservoir and modeling software packages. Once all configurations have been set, succession attribute trends are computed within minutes, and, beyond computed property statistics, new factors describing the heterogeneity of the interval of interest are introduced. The user can, for example, estimate the likeliness of two sand beds being vertically connected or how heterogeneous reservoirs or units in general are developed. It therefore lends itself to support predictions on sediment transport fairways as well as to determine trends in dipping or faulting – all in a fully quantitative format. The entire workflow has the ability to reproduce with more than 80% fidelity against calibrated core and reduces the time spent for static basin model building by ~ 35–45%.