Linking compositional data analysis with thermodynamic geochemical modeling: Oilfield brines from the Permian Basin, USA

Compositional data analysis (CoDa) of sedimentary basin brines can help reveal the hydrogeochemical history of hydrocarbon producing formations. CoDa uses log-ratio transformations to convert compositional data into Euclidean space, a necessary assumption for many mathematical models and basic data interpretation. This paper demonstrates the novel approach of pairing CoDa with thermodynamic geochemical models for bivariate and multivariate subcompositions for a suite of brines within three producing formations from the Permian Basin, USA. Subcompositional analysis of basin brines using [Na, Cl, H2O] and [Ca, SO4, H2O] with corresponding equilibration models, plotted using isometric log-ratio (ilr) transformations, shows reaction with anhydrite and halite at depth. The [Na, Cl, Br] subcomposition often can discriminate between seawater evaporation and halite dissolution; the sparse Br data suggest that both processes occur. Principal component analysis (PCA) of centered log-ratio (clr) data is useful for visualizing combinations of processes that are not readily apparent in the two-solute models. PCA of the Permian Basin brines reveals distinct groups of data, including samples likely affected by water flooding, and others influenced by a combination of Late Permian seawater evaporation or halite dissolution coupled with Ca + Mg ion exchange for Na on clays. These relationships support a hydrogeologic model in which meteoric inflow occurs in a heterogeneous manner much more complicated than the previously suggested concept of west-to-east piston style flow.