Technology and Activity Introduction
- Digital Technology -

Digital Technology Development

JOGMEC is challenging to improve the efficiency and reduce uncertainty of resource development business using new ideas of digital technology, such as AI. Digital Transformation Group is implementing studies from a PoC (proof of concept) to field demonstrations using digital technologies for solving technical problems faced by each Japanese company.

Automatic Carbonate Lithology Identification

A:Microscope image of carbonate rock
B:Analytic image interpreted by an expert
C:Analytic image interpreted by AI
(Nanjo and Tanaka, 2019 )

Automatic Carbonate Lithology Identification

We are developing our own deep learning model, which conducts carbonate lithology identification. Carbonate rocks are the main reservoirs of giant oil and gas fields in the Middle East. Our model is currently able to conduct carbonate lithology identification with over 90% accuracy. Our goal is to increase its versatility, such that it can be used in various oil and gas fields.

Automatic Calcareous Nannofossil Identification

Left:Example of a microscopic image of calcareous nannofossil
Right:Microscopic image of a sample containing calcaleous nanofossil (the object in the red frame is a calcareous nannofossil)

Automatic Calcareous Nannofossil Identification

The age of the formation currently being penetrated can be evaluated by identifying fossils contained in the drilling cuttings. However, there are a limited number of specialists with the requisite knowledge for fossil identification. In collaboration with industry and academia, we are developing an image recognition AI model that can identify fossils.

Automated Fault Interpretation

Example of faults observed from outcrops

Automated Fault Interpretation

Properly interpreting the position of faults from seismic data is useful in preventing drilling problems and preparation for an appropriate oil and gas field development plan. However, variability in fault position interpretation occurs based on the interpreters. Therefore, we are testing various methods to rapidly, accurately, and objectively interpret faults positions using machine learning.

Automatic Identification of Reservoir Properties and Facies

Example of lithology prediction result from machine learning model by analyzing geological data

Automatic Identification of Reservoir Properties and Facies

Specialists evaluate subsurface oil and gas reservoir properties and facies by analyzing geological data obtained from wells. We are developing technologies that enable this evaluation through machine learning model instead of the specialists, with the goal of improving estimation accuracy.

Improving Safety Performance in Drilling Operation

Example of a drilling trouble (pipe sticking)

Improving Safety Performance
in Drilling Operation

Drilling operation may be interrupted when the drilling pipe is stuck with formation and becomes immobile. We are developing an AI that can detect signs of such problems in collaboration with private companies/research institutions with knowledge on well excavation, and domestic and international universities with knowledge on AI development.

Development and Production Plan Optimization

Example of reservoir simulation of CO2-EOR

Development and Production Plan Optimization

We build oil/gas reservoir models to match oil/gas production history, and to predict future oil/gas production. In order to optimize the model, we’re developing data analysis technologies. And we’re also developing the methods to predict oil/gas production without reservoir models, which analyzes the relationship between water injection rate history and oil production rate history.

LNG Value Chain Optimization

Example of an LNG vessel

LNG Value Chain Optimization

In the LNG business, the efficient combination of LNG vessel delivery routes and LNG vessel sizes must be determined from an astronomical number of candidates. We demonstrated that these calculations, reputedly difficult even for a supercomputer, can be conducted with quantum computing technologies.

HRD and Technology Trend

Our goal is to be the group with expertise in both resource development and digital technology.
In order to achieve this goal, group members with backgrounds in technologies related to resource development are engaged in acquiring techniques, such as machine learning or deep learning. We aim to enable members themselves to play a leading role in the utilization of digital technologies.
We also aim to catch up with the rapidly evolving digital technologies. We analyze and offer the latest trend information to the entire resource development industry in Japan in order to enhance their technical level.

A:Our goal of human resource development is to cultivate personnel with specialty as shown in this figure.
B:We are using JDLA certification as a progress milestone of the human resource development.
(Note) JDLA : Japan Deep Learning Association

DX of JOGMEC's Business Workflow

JOGMEC is proceeding with digitalization of our own business workflow in order to provide quick and convenient services for those who use our support system. We are striving for optimization of our business workflows from a digital perspective to make it beyond mere digitization of existing workflow in analog.