Geostatistical facies modelling of cyclicity and rhythmicity in shallow-marine reservoirs

Abstract

This work presents a new geostatistical approach aimed at incorporating more accurate geological input into existing Earth modelling tools. These tools play a key role in petroleum reservoir and mining deposit modelling workflows as they are the basis for the estimation of resource recovery and its associated uncertainty. Conventional geostatistics generate facies successions that are statistically identical along one direction or its opposite. This impedes the modelling of facies ordering, also called facies cyclicity, which is often characterized in shallow-marine carbonate and siliciclastic reservoirs by shallowing upward facies cycles. Cyclicity is often coupled with another geological characteristic of sedimentary sequences, which is that facies usually appear repeatedly at almost constant stratigraphic intervals, a feature called facies rhythmicity. The principal result of this thesis is a new method that can quantify and simultaneously model facies cyclicity and rhythmicity. Cyclicity is quantified using transition probability analysis, while rhythmicity is quantified using two point statistical tools such as transiograms. The new method is an extension of the approach known as Pluri-Gaussian Simulations, which is significantly improved by developing new periodic covariances and a co-regionalization model with spatial asymmetry. The method was tested on case studies of shallow marine deposits such as the Latemar carbonate platform in Italy and the Blackhawk Formation siliciclastic shoreface deposits in the Book Cliffs, USA. The choice of case studies was dictated by the desire to model a range of geological patterns including different styles of diagenesis, the presence of erosional surfaces, and non-stationarity. This was also an opportunity to evaluate the method in relation with established geological concepts such as sequence stratigraphy or Walther's Law. By generating more realistic three-dimensional geological Earth models, the method has the potential to significantly improve the quality of hydrocarbon and mining resource estimates and the quantification of their associated uncertainty.