Enhanced Reservoir Characterization Using Machine Learning

Abstract

This study has proposed a novel approach for improving reservoir characterization by compartmentalizing the interval into subsections with highest consistency in pore-throat/body size and distribution by means of intelligent machine learning. Hydraulic flow units have demonstrated success in segmenting the interval of interest into subsections with distinguishable rock and fluid properties. Firstly, flow zone indicator values are calculated from core data within a reservoir of interest. Flow zone indicator is an acceptably unique measurement of flow character of a reservoir interval offering a relationship between petrophysical properties in pore scale, like tortuosity and surface area, and formation scale, say porosity and permeability. The inputs are primarily analysed based on principal component analysis to be accurate representatives of the underlying compartments. Next step segments the reservoir into more accurately delineated intervals, hydraulic flow units, whose number is determined using the statistical approaches. Several statistical approaches have demonstrated success in grouping data into subsections of high similarity and consistency, herein hydraulic flow units. The robust approach proposed in this paper can lead to thriving in reservoir management by saving considerable amount of revenue merely by accurately predicting interval flow properties without the need for expensive coring operation, based on well log data.