Successful Identification of Bypassed Pay Zones and Wet Sands, in a Mature Oil Field, Using Multi-Frequency Dielectric Logging and Advanced Radial Inversion Methods: A Case Study From Colombia

Abstract

This paper presents a case study from a mature oil field subjected to long term injection of fresh water. Multiple, potential oil horizons are encountered in newly drilled wells, however, formation evaluation is very challenging due to mineralogical heterogeneity and mixing of formation and contrasting injected waters altering formation water salinity to variable levels. Conventional logs fail to accurately predict oil/water saturation and reservoir quality under these conditions. Multifrequency dielectric measurements were acquired and inverted using a newly developed radial inversion method, to establish the length of invasion and complex permittivity of both invaded and uninvaded formation layers. Using this method saturations are determined, independent of salinity, and are successfully tied to mineralogy related reservoir quality variations. A multi-frequency, multi-spacing, array dielectric instrument measures formation permittivity with various radial depths of investigation. These high frequency measurements require closely spaced receivers also providing measurements of thin layers at high vertical resolution. An advanced radial model inversion algorithm was developed, which employs a trained, deep-learning artificial neural network (ANN) to invert the large quantity of data acquired. The algorithm resolves the dielectric properties of inner/invaded and outer/uninvaded formation layers. In a subsequent step, the frequency dispersion profile is inverted separately for these two regions in order to determine water saturation, salinity and textural parameters. Complex reservoir structure, variations in mineralogy, long term injection of fresh water for secondary recovery, and use of fresh drilling mud makes it highly challenging to interpret the response of conventional log measurements, leading to uncertainty in identification and assessment of pay zones. The new methodology was successfully applied in this case history to resolve these uncertainties. The study well was drilled with a mud salinity much lower than formation salinity. In zones with shallow mud filtrate invasion, dielectric measurements extend over an area of fluid heterogeneity, from invaded to virgin formation, and a simple homogeneous formation model is not representative. The advanced model considers radial layers of borehole, mud-cake, invaded and uninvaded formation, and correctly inverts for variable formation properties away from the wellbore. Petrophysical results show correct identification and quantification of prospective oil zones well differentiated from wet sands. Independent evaluation of mineralogy is integrated to further understand the saturation variations in productive layers. The information that this technique supplies is of immense benefit to petrophysicists, completion and reservoir engineers engaged in development of mature fields using waterfloods.