Reservoir Fluid Properties From Cuttings: An Innovative Synergy of Gel Permeation Chromatography and Data Analytics

Drilling cuttings, often deployed for lithology and mineralogy analyses, hold untapped potential. Extracts from cuttings are valuable in traditional geochemical analysis, particularly in water-based mud (WBM) applications. On the other hand, the scarcity of reservoir fluid samples from reservoir zones and overburden presents a challenge, withholding vital insights crucial for well integrity, plugging and abandonment (P&A), and efficient reservoir management and production. Paradoxically, numerous drill-cutting samples remain unexamined within storage facilities. We have achieved a significant milestone after an intensive two-year research effort focused on developing an innovative gel permeation chromatography (GPC) technique for analyzing reservoir oils and cutting extracts. Our innovation facilitates the in-depth examination of reservoir fluid properties, i.e., API gravity and viscosity, from oils and cutting samples, effectively overcoming challenges associated with oil-based mud (OBM) contamination. In this study, we investigate the application of GPC coupled with both ultraviolet (UV) and refractive index (RI) detectors to generate multidetector spectra from reservoir fluids and cuttings extracts originating from six distinct fields situated in the Norwegian Continental Shelf (NCS). The technique is deployed to analyze oil samples, encompassing a wide range of reservoir fluid types, including condensates, volatile oils, black oils, and heavy oils. The primary goal of the study is to estimate reservoir oil density using GPC-UV-RI spectra from cuttings. Following acquiring GPC spectra, the data are compiled into a vectorized data set for subsequent processing in a data analytics workflow. This data-processing phase comprises exploratory data analysis and quality checking, data augmentation, and machine-learning modeling that serves as a proof of concept of the application. A suite of machine-learning algorithms, including regularized linear regression models, are evaluated, and a comprehensive comparison of performance, generalization capability, and robustness of the baseline and augmented models are also discussed. The developed models demonstrate robust accuracy in predicting reservoir fluid properties, specifically °API (American Petroleum Institute gravity). Fundamentally, this technology has large potential to transform each drill-cutting fragment into a PVT (pressure-volume-temperature) sample and contributes significantly to unlocking the latent value within the cuttings, which are readily available and offer access to reservoir fluid properties at an early stage in field development, including the drilling phase. The application of the technology is not only cost effective but also carries far-reaching implications that extend into various areas, such as strategic well placement, enhanced wellbore integrity, and optimized completion strategies. This innovation marks a significant step towards maximizing value creation in exploration and production operations.