Today, the major challenge in reservoir management is improving reservoir characterization to better understand variations in rock properties and distribution of fluids away from the wells. A further challenge is the proper characterziation of thin multi-layered heterogeneous reservoirs. In this paper we present a new workflow to delineate reservoirs at and below seismic resolution.

A reservoir characterization study is performed for the Reservoir Z of the Giant field. The reservoir seqeuce consists of multiple reservoir units (1, 2, and 3) with variable average porosities between 11 and 28%. The study attempts to delineate separate units through an innovative Bayesian inversion technique that jointly solves for impedances and facies (Kemper and Gunning, 2014). Four different seismic facies were determined through an inversion feasibility study from variations in mineralogy and porosity. Elastic property trends as a function of time were built from the petrophysics of two wells. The low frequency component was driven by these trends and prior facies distributions were specified to constrain the results to be geologically reasonable. Three angle stacks were simultaneously inverted using extracted wavelets, each with specified noise estimates. An inversion was run using this technique on post stack data, as well as a coloured inversion for comparison.

Despite varaible data quality, the pre-stack facies based Bayesian inversion was able to invert for three separate layers of high porosity in the reservoir sequence in a number of blind wells. This was a noteable improvement on both the coloured inversion and post-stack inversion. It was concluded that the higher frequenices in the near stack data combined with increased rock physics contraints of the inversion resulted in better thin layer detection.

You can access this article if you purchase or spend a download.