A live test of automated facies prediction at wells for CO₂ storage projects

Abstract

At least 900 m of image log data have been interpreted in detail through the Late Cretaceous and Cenozoic succession intersected at two purpose-drilled injection wells of the CO2CRC Otway Project. Interpretations have been calibrated against core observations where possible. Natural clusters for the combined signature of a common suite of coincident well log data were determined using an unsupervised Naive Bayesian classification algorithm called Autoclass. A deterministic relationship between these modelled clusters and interpreted image log facies provides a two-step facies prediction algorithm that can be applied using well log data acquired at other wells intersecting sedimentary successions prospective for CO2 storage. Earlier this year the Division of Resources & Energy, Department of Trade & Investment, of the Australian state of New South Wales drilled a new stratigraphic test well within the Pondie Range Trough of the Darling Basin. The new well, Mena Murtee-1, was drilled close to a 2D seismic tie line linking the Pondie Range Trough depocenter with an old petroleum exploration well, Pondie Range-1, that is sited on a flanking high. Analyses and interpretation of data acquired at Mena Murtee-1 is the latest step in reducing the uncertainty surrounding CO2 storage potential within the Darling Basin. Interpretation of core and image log data acquired at the new well has provided a means by which to test facies predictions made on the basis of the models developed within the Otway Basin.