The investments in carbon capture and sequestration (CCS) projects are steadily increasing in many parts of the world. Despite the more common use of aquifers, depleted oil and gas reservoirs are appealing storage targets because of the existing subsurface knowledge and surface infrastructures expected to facilitate the process. However, existing lithological and petrophysical models designed for oil and gas may not necessary be CCS-ready.
The paper describes the challenges and the solutions developed for accurate mineralogical characterization of depleted gas reservoirs selected for CO2 injection and their caprock, which is essential to determine the storage capacity of the asset and to provide an accurate baseline to modelling and simulation through the project.
The formation of interest in this paper is characterized by mixed lithology, with minerals including quartz, feldspars, micas, carbonates, and different clay types. Existing mineral models based on conventional openhole logs collected during exploration and appraisal are not sufficient to solve for the detailed mineralogy needed for CCS, and they do not extend to the sealing layers. Recent elemental spectroscopy measurements from advanced pulsed neutron logging performed in cased holes confirmed that specific mineral groups seen by core data are of significant occurrence downhole.
Extended analysis of existing core data and important knowledge of rock elemental composition and mineral variations gathered during cased hole evaluation are leveraged to re-interpret the existing petrophysical reservoir models and to characterize the non-reservoir units. Two methodologies based on core-log integration are discussed; their results can be compared for consistency and additional understanding. Both approaches start from a detailed scrutiny and analysis of core data to develop a locally optimized continuous mineralogy from logs.
The first approach, a data analytics physics-based method, already available (Pirrone et al, 2022) uses integrated data-driven analytics to generate synthetic volumetric fractions of given minerals, with a result that relates core mineralogy to selected logs. Those mineral fractions are then combined with other openhole logs into a multi-mineral solver.
The second approach is a novel core-to-log workflow using dimensionality reduction techniques applied to core data to capture the inherent correlations between minerals. Different mineral assemblages are described, and the learning is extended to the definition of associated end points for cased hole and openhole logs, subsequently used within a multi-mineral solver.
The data-drive analytics-informed physics-based method can have a primary role in re-defining the petrophysical model based on the large database of openhole logs across the field, from single to multi-well coherent analysis. The novel core-to-log workflow provides an alternative solution and critical addition in cases where openhole logs are missing or limited, and the evaluation is based on cased hole logs performed in completed wells. Depending on the complexity and available pulsed neutron logs, more elaborated or more simplified mineral assemblages can be selected to build a simplified yet fully calibrated lithological and petrophysical model.
