Building a Predictive Model in a Fluvial, Low Net-To-Gross, Multilayered Reservoir With Heavy Oil: A Case Study From El Guadal Sur (EGS) Field, Golfo San Jorge Basin, Argentina

Abstract

The western flank of the Golfo San Jorge Basin is home to a number of oil fields and thousands of wells, yet recovery factors are generally very low. Water-flooding has had a highly variable degree of success, probably due to a poor understanding of the reservoirs. These were traditionally interpreted and correlated as isolated sandstone bodies with very limited lateral extension and connectivity, and treated as internally homogenous. Correlations were highly interpreter-driven, and there was no hierarchy of sandstone bodies. Hence, the main purpose of this study was to understand and optimize the current water-flooding, and identify future opportunities. EGS is an anticline structure segmented by minor faults. The main reservoirs are the Cretaceous fluvial sandstone bodies of the Bajo Barreal Formation, deposited in a post rift setting influenced by volcanic activity. Reservoir characterization is challenging, sandstone bodies being only 2–5m thick, heterogeneous, finely interbedded with clay in an overall column thickness of 700m. Seismic resolution is poor, well log data is abundant but restricted to spontaneous potential, resistivity and occasionally density logs, core data is scarce, and a lack of PLT data makes 3D static and dynamic simulation the only way to estimate which sandstone bodies are actually being produced. A new methodology (Vertical Proportion Curve, VPC) was developed for correlation, aiming at reducing dependence on interpreter's criteria, and based on averaging log data at the same interval from groups of wells belonging to the same fault block. The VPC clearly shows the most continuous shaly intervals (minimum energy in the system) used to subdivide the unit, and the higher net-to-gross (NTG) zones that will respond better to a water-flood. Based on this framework, successive 3D models were built, integrating the available seismic, outcrop, core and log data. Modelling effort was focused on a realistic representation of fluvial geometries, applying either Object Modeling or Multipoint Statistics. Different scenarios were tested, going back and forth with the dynamic simulation. The results from modelling show that connectivity is required to match the production history, and that most of the oil comes from a few intervals with higher NTG. As a result of this study, the understanding of the production hierarchies has enabled a number of improved oil recovery schemes to be considered, and possibly a polymer pilot.

AAPG Datapages/Search and Discovery Article #90216 ©2015 AAPG Annual Convention and Exhibition, Denver, CO., May 31 - June 3, 2015