Outcrop-Based Fluvial and Lacustrine Sandstone Reservoir Characterization and Modeling of the Eocene Uinta and Duchene River Formations, Northern Uinta Basin, Utah
Over the last two decades, stochastic reservoir modeling techniques have progressed to the point where they are now commonplace in asset team workflows performing reserve estimates and recovery predictions. While there are multiple approaches for building reservoir models which can generate literally thousands of different realizations constrained to a limited number of control wells, outcrop analogs provide an invaluable groundtruth to guide reservoir modeling for the most realistic outcomes. Fluvial and lacustrine deposits in the Eocene Uinta and Duchesne River Formations of northeastern Utah represent the late stage filling of the intermontane Uinta Basin. These formations are well exposed at Blacktail Mountain in the western part of the basin, revealing a vertical cliff-face section of 206 m with 1856 m lateral exposure. Prominent cyclic, prograding deltaic sequences are overlain by fluvial channel dominated sequences. A high resolution GigaPan photo and four vertical measured sections were used to interpret the detailed facies architecture. These field data were also compared to subsurface well logs covering the same stratigraphic levels in a 30 km west-east section. The combination of data sets provided statistics of fluvial channel geometry and distribution. Facies interpretations of the Blacktail cliff-face were translated into a pixel-based geologic reference model with ~120,000 cells (dimensions dx = 10 m, dy = 10 m, dz = 1 m). Facies were manually assigned to each cell. This reference model of the fluvial unit was compared to reservoir models generated by three techniques: 1) indicator kriging, 2) sequential indicator, and 3) object-based modeling. All models were constrained by: a) two measured sections as hard data, b) global facies proportions of 50% fluvial channel and 50% flood plain, and c) channel geometries (variogram or width:depth statistics). A static connectivity analysis determined which modeling technique most closely reproduced the static sandbody connectivity characteristics of the reference outcrop model. This analysis showed that object-based models fit the reference model best in terms of wellbore connectivity, although some realizations showed large deviations from the reference case. Overall, this study provides an important analog example that leverages statistical inputs in geological modeling, and demonstrates which stochastic modeling techniques best represent observed depositional patterns derived from outcrop data.
AAPG Datapages/Search and Discovery Article #90216 ©2015 AAPG Annual Convention and Exhibition, Denver, CO., May 31 - June 3, 2015