Routine core analysis (RCA) for Reservoir Quality (RQ) prediction comprises a collection of measurements acquired with different analytical techniques, which does not include operator-bias evaluation, or integration of continuous sedimentological core description with spot measurements on plugs and thin sections. RCA data rarely have verified uncertainty specifications, thus hampering statistically-rigorous extrapolation of spot measurements such as petrographic description, to the entire reservoir volume. Petrographic analysis gives insight into the controls on RQ through unravelling the diagenetic fingerprint that shapes the eventual porosity and permeability in the reservoir. Because thin-section analysis is time consuming and costly, protocols for selection of representative thin sections should aim at maximizing information obtained from small data sets, so as to minimize costs and prevent unnecessary destruction of core material. This paper presents a flexible protocol for representative thin-section selection based on evaluation of RCA data (i.e., poro-perm and grain-density plug measurements), illustrated on a core of a Carboniferous fluvial sandstone reservoir. The results of the petrographic analysis are interpreted in terms of their relation with sedimentological and geochemical signatures, and it is demonstrated that application of the protocol highly increased the RCA data value which to date merely served as petrophysical indicators.@en

In light of the huge investments needed to acquire sediment cores and the growing need of energy-providing companies to predict reservoir quality, it is remarkable that the standard workflow in core analysis has not been optimized to extract as much information from cores as possible. The Integrated Core Analysis (ICA) protocol presented in this study provides a statistical framework for multivariate calibration and prediction of a wide range of properties measured in core, based on integration of high-resolution X-ray fluorescence core-scanning (XRF-CS) proxy records with records of sparsely sampled petrophysical data and sedimentological core descriptions. The downscaling of numerical data involves the application of invertible transformations to remove range constraints on data, followed by calibration with Partial Least Squares regression through cross validation. Categorical data (i.e. lithofacies) are downscaled by associating each class with a statistical model based on the XRF-CS data. All data points are reassigned to their most likely class by applying Quadratic Discriminant Analysis. The result of the ICA is a multivariate data set in which all properties are specified at the same high resolution along the core with their prediction uncertainties. Downscaling of all variables to 1-cm vertical resolution permits investigation of the variability among petrophysical properties, geochemical proxies, and lithofacies memberships. Petrographic analysis is fundamental for interpretation of the XRF-CS records (element-mineral affinity) and for understanding the sedimentological controls on predicted petrophysical properties. Application of the ICA protocol to a 32 m thick, heterogeneous, Upper Carboniferous fluvial sandstone interval resulted in a near 30-fold increase of the petrophysical data base, which allowed identification of the main depositional and diagenetic controls on the spatial distribution of reservoir quality. Successful implementation of the novel ICA protocol will greatly increase the economic value of legacy core data in studies that aim to re-use depleted hydrocarbon reservoirs.@en