Reservoir compartment prediction of the Simpson field from the geostatistical inversion of AVO seismic data

Abstract

The Simpson field in the Barrow Sub-basin (Carnarvon Basin) is nearing depletion. Most of the producing wells are showing relatively high water cuts. Based on volumetric mapping and the drilling results from nearby analogous fields, some unproduced reserves are potentially remaining in the field within compartments separated by low permeability shale barriers. The challenge is to establish a methodology for identifying these compartments and to quantify unproduced oil. The reservoir of the Simpson field is the Early Cretaceous Flag Sandstone. The reservoir zone has three distinct lithotypes: oil-saturated sandstone, water-saturated sandstone and shale. The shales encountered in the wells have a typical thickness of less than 3 m, significantly below seismic resolution. However, these lithotypes show good statistical separation of elastic properties (e.g. P-impedance and Vp/Vs), so a properly-constrained geostatisitcal inversion can be used to predict the relatively thin shale barriers. The geostatisitcal inversion is based on Bayesian Inference and uses a Markov Chain Monte Carlo method to randomly sample the probabilistic distribution arising from all input beliefs. The end result is multiple elastic property and lithotype realizations. The analysis of the realizations output from the geostatistical inversion led to improved reservoir volumetric calculations and the identification of highly probable shale barriers.