Dolomite is a commonly found mineral on earth; it plays a significant important role in making up hydrocarbon reservoirs.

Dolomite has been the focus of academic study for a long time, but its genesis is still largely unknown because experimental studies to produce dolomite at room temperature and pressure have not been successful.

Multiple models have been proposed for the formation process of dolomite; the accuracy in evaluating a carbonate reservoir property depends on the model that is selected. Predicting dolomite reservoir property is therefore a big challenge in petroleum field development.

INPEX is focusing on the dolomite that forms reservoir rocks in a giant oil field offshore Abu Dhabi, United Arab Emirates. To determine the formation mechanism of these reservoir rocks, we are working together with a Japanese university for combining their academic expertise in geochemical analysis with our specialty in reservoir rock characterization.

Through this study, we aim to attain an efficient recovery of hydrocarbons by optimizing the location and design of wells based on the insight into the spatial distribution of the dolomite reservoir.

When interpreting underground geology, available information is often limited in quantity and quality. Exposed outcrops on the surface are referred to as valuable analogues to assess rock properties and spatial extent of subsurface reservoirs in oil and gas fields. These outcrops are also used as teaching materials for engineers and geoscientists who are involved in subsurface evaluation to deepen their understanding of a reservoir and to share geological concepts among them.

In 2022, INPEX conducted a geological survey by employing drone photography to effectively collect detailed information on lithofacies from outcrops. While conducting the aerial photography with a drone, we measured geographical coordinates of the ground control points installed at outcrops so that we could restore the georeferenced 3-D digital models of outcrops precisely. The built outcrop models enabled us to observe and interpret sedimentary structures and textures of the rocks exposed on cliffs that are inaccessible. Digitally preserved shapes and colors of outcrops allowed us to add rock descriptions and collect geological information without revisiting the site.

INPEX aims to utilize the created 3-D digital outcrop models not only for the improvement of reservoir modeling but also for training employees.

Recent advances in technology have made it possible to obtain a substantial amount of relatively high-quality information for subsurface geology. However, the available information remains limited in quantity, and much of manpower and cost are being spent for the improvement of subsurface reservoir models and mitigation of their uncertainties.

INPEX has been developing methodologies that employ machine learning, which has displayed remarkable advances in recent years. One example of such a technique is seismic inversion, which inversely analyzes the elastic property of subsurface rocks using seismic data. Another example is improvement of the quantitative seismic interpretation techniques, which analyze lithofacies and reservoir rock properties based on the results of seismic inversion.

In conventional seismic inversion, a model is manually built from existing wells to compensate low-frequency components missing in seismic data. This approach results in a wide range of flexibility in creating the models at locations away from wells and a high dependency of the inversion analysis on these models. Quantitative seismic interpretation is based on direct relationships among elastic properties, lithofacies, and reservoir properties, which makes it challenging to incorporate known subsurface information like occurrence patterns of lithofacies into analysis. INPEX has been developing novel analysis techniques that reduce the dependence on low-frequency models and allow the integration of known subsurface information by introducing machine learning.

The locations and sizes of faults need to be determined in the exploration and development of oil and gas, because faults strongly influence the geological interpretation and may emerge as potential risks for drilling.
Notably, seismic reflection survey is the most popular method for acquiring the information of underground geological structures, which uses artificially generated ground motion. Accordingly, we determine the existence of geological boundaries and faults by interpreting the cross-sections of geological structures, primarily through manual interpretation by specialists.
Depending on the sizes of survey areas, seismic interpretation generally requires a few weeks to a few months. In addition, we encounter certain issues while engaging multiple specialists. The interpretation quality may not be consistent as the interpretations are based on individual perspectives and experience.

To resolve these issues, efficient and highly accurate automatic interpretation utilizing machine learning has been introduced for seismic interpretation. Consequently, in 2019 we achieved the ability to automatically interpret faults and conduct efficient interpretation work to a certain level of accuracy, and we further enhanced this ability to the level of practical application as of 2022. Thus, INPEX will continue to invest efforts in improving the efficiency and accuracy of geological evaluation, including automatic interpretation.