Methods for predicting mineralogy from logging tools measurements have been an active area of research for several decades. In spite of these efforts, methods for predicting quantitative mineralogy including clay types from well logging data were not fully achieved. The introduction of geochemical logging tools in the 1980s offered promise; however, early versions of geochemical logging tools did not measure elemental chemistry with enough accuracy and precision to enable reliable and quantitative determination of mineralogy. Recent advances in geochemical logging tool technology now enable accurate and robust measurements of the chemical elemental concentrations that are needed to determine continuous quantitative and detailed logs of mineralogy.

This paper presents a novel approach for determining more accurate, detailed mineralogy from an elemental spectroscopy logging tool. This work was made possible by three recent developments: the introduction of a new high-performance neutron-induced gamma ray spectroscopy logging tool, a new high-quality research database consisting of chemistry and mineralogy measured on cores acquired worldwide from conventional and unconventional reservoirs, and a new model-independent inversion method that overcomes the limitations of previous model-dependent methods.

The model-independent inversion makes use of the database which includes clean sands, shaly sands, shales, carbonates, and complex mixed lithologies. The database contains laboratory measurements of dry-weight elemental chemistry and mineralogy measured by transmisson Fourier transform infrared spectroscopy. The database is used to derive a model-independent mapping function that accurately represents the complex functional relationship between the elemental concentrations and the mineral concentrations. Once the mapping function is determined from the database, it can be used to predict quantitative mineralogy from elemental concentrations derived from the logging tool measurements. Unlike previous inversion methods, the model-independent mapping function does not have any adjustable parameters or require any user inputs such as mineral properties or endpoints.

The mapping function is used to predict continuous logs of matrix densities plus concentrations of 14 minerals (i.e., illite, smectite, kaolinite, chlorite, quartz, calcite, dolomite, ankerite, plagioclase, orthoclase, mica, pyrite, siderite, and anhydrite) from eight dry-weight elemental concentrations derived from the logging tool. The new method has been applied to well log data acquired worldwide in numerous conventional and unconventional reservoirs having a wide variety of complex mineralogies. The predicted mineralogies and matrix densities are generally found to be in good quantitative agreement with core-derived mineralogies and matrix densities.