The Application of Very Large Numbers of Pseudo-Wells for Reservoir Characterisation

Abstract

We describe a method to invert seismic data by matching to large numbers of pseudo-wells. Pseudo-well construction is based on lithofacies micro-layering constrained by a model for the vertical statistics. Elastic properties are derived from a rock physics model. We match either to single or multiple angle-stacks using synthetics based on extended elastic impedance. This is a joint inversion process as it simultaneously accounts for both elastic parameters and reservoir properties. Both localised means and standard deviations of reservoir properties are estimated. We do not include priors for lateral statistics but instead we envisage this process forming the first step of a two stage workflow with the output providing constrains for geostatistical modelling in the second stage. With no constraints on lateral correlation the first stage can operate in 1D greatly decreasing computational requirements. We show examples of application to two fields; one offshore Angola and the other offshore West of Shetlands. These demonstrate the quality of the algorithm which provides laterally stable results, high signal-to-noise modelling of the seismic and accurate blind well predictions.