Challenges in characterizing subsurface paralic reservoir geometries: a detailed case study of the Mungaroo Formation, North West Shelf, Australia
G. Heldreich, J. Redfern, B. Legler, K. Gerdes, B. P. J. Williams, 2017. "Challenges in characterizing subsurface paralic reservoir geometries: a detailed case study of the Mungaroo Formation, North West Shelf, Australia", Sedimentology of Paralic Reservoirs: Recent Advances, G. J. Hampson, A. D. Reynolds, B. Kostic, M. R. Wells
Download citation file:
Prediction of sandstone body dimensions within paralic depositional systems is crucial for the development of predictive 3D reservoir models. Continental-scale paralic reservoir targets have complex architectures, with interpretation often further compounded because they are often located at subsurface depths ≥2 km in pre- and synrift basinal settings, typified by poor seismic resolution. As such, in many cases analysis relies on core and wireline-log data, from which depositional facies are interpreted and thicknesses of sandstone reservoir units measured. Estimation of width:thickness (W:T) ratios for different reservoir elements relies on analogue data. However, the inherent uncertainty in the initial interpretation of core/log data and the wide W:T ranges for different sandstone bodies within published analogue datasets hinders prediction of accurate reservoir geometry/dimensions. Without access to quality seismic data, constraining the evolution and dimensions of reservoirs in these depositional systems is challenging.
This paper presents a detailed case study of the Triassic Mungaroo Formation, and assesses the uncertainty and limitation of interpretations that can be made about reservoir architecture and the evolution of a paralic depositional system, if only wireline and core data are available.
The Mungaroo Formation is characterized by upper and lower delta-plain channel sandstones, swamps and restricted embayments, through to delta-front and pro-delta heterolithics. Core-to-wireline-log calibration allowed identification of key marine intervals that enabled well correlations to be established across the study area, based on candidate flooding surfaces. Applying classic sequence stratigraphic models to low gradient fluvio-deltaic systems is difficult, due to a lack of preservation and/or confident identification of laterally continuous chronostratigraphic markers. The large-scale temporal change in reservoir architecture was likely to have been controlled by eustatic sea level causing overall transgression of the depositional system. However, the observed complex spatial facies variability is most likely controlled by climatic-induced changes in sediment supply/fluvial discharge and autocyclic processes.
Four classes of sandstone bodies/reservoir elements have been identified from core and wireline data based on their thickness distributions. Whereas core to wireline to seismic calibration has enabled large-, medium- and small-scale geobodies to be identified representing fluvial channel belt complexes; multistorey channel belts; and single-storey channel belts, respectively. To predict channel body widths and sinuosity from thickness data extracted from wells, typically, analogue data are used. However, for each geobody type there is a large range of possible sandstone body dimensions based on published literature. The largest-scale sand bodies, with most significance as reservoir targets, have possible interpretations as either incised valleys or amalgamated channel belts, based on their thickness ranges alone. This poses significant uncertainty for understanding the evolution of the depositional system and input into predictive reservoir models.
This study emphasizes the importance of understanding the range of uncertainty of interpretation and the need for refined analogue data to better constrain reservoir element dimensions when relying solely on well-log datasets. Where seismic attribute analysis from high-quality 3D seismic data are available, W:T dimensions for reservoir elements can be constrained more accurately and correlated to core and log data. The existing global database is limited, often poorly constrained due to the use of variable terminology and potential for misinterpretation. Many studies lack statistical rigour. We conclude that further high-resolution studies are required to build more robust and quantitative analogue datasets.