Idemitsu Kosan has been a cornerstone of Japan’s energy supply for over 100 years since its founding in 1911. The company develops, manufactures, and sells a wide range of energy and materials, including fuel oil, basic chemicals, high-performance materials, electric power, renewable energy, and resources.
The company's Technology & Engineering Center supports the Manufacturing Engineering Sector in driving structural reform of core businesses and shifting its portfolio toward new ventures. Guided by the mission of "creating new value through wisdom and technology and fulfilling our responsibility to implement this in society," the center is leading the company’s digital transformation (DX) initiatives.
As part of these efforts, the center adopted Cognite Data Fusion® in April 2024 to create a virtual integrated database that enables data integration and utilization. By November of the same year, Cognite Data Fusion® was deployed at four of Idemitsu Kosan's manufacturing sites that focus on petroleum refining and chemicals. To gain insights into this transformative journey, we interviewed Naruki Akiyama and Hanako Kubo, who spearheaded the implementation of Cognite Data Fusion®.
Issues prior to Cognite Data Fusion® implementation
For years, Idemitsu Kosan struggled with data silos. Since the 1990s, equipment records, operational histories, handover notes, and accident reports have been digitized gradually. However, these systems were fragmented, and their data needed to be more interconnected. Maintenance planning often requires manually collecting data from various systems—ERP, operational data (PIMS), equipment libraries, and file servers—and organizing it using tools like Excel. This process consumed a significant amount of time and resources.
By creating a centralized data platform, the DX initiative aimed to streamline access to disparate information, drastically reducing the time required for data collection and improving overall productivity. Around 2018, the company began exploring the concept of a virtual integrated database to manage data centrally.
Background from the start of the study to its introduction
At first, the idea of a virtual integrated database faced resistance. However, in 2020, Idemitsu Kosan discovered Cognite and began technical evaluations. The results highlighted the superiority of Cognite’s contextualization technology. Features such as AI-powered Optical Character Recognition (AI-OCR) for extracting text from images and drawings and automated linking of ambiguous data keys enabled efficient data integration in a fraction of the time manual efforts would require.
While the technical evaluation was straightforward, convincing departments to adopt the system posed challenges. They worked to demonstrate how Cognite Data Fusion® could integrate with existing systems and improve operational workflows. By mapping out the system’s potential impact, they developed a comprehensive roadmap for integrating the database with the Manufacturing Engineering Sector's systems.
In October 2023, a cross-departmental feasibility study was launched. Six use cases were created using feedback from the manufacturing site on Chiba Complex, and real-world data was partially linked to Cognite Data Fusion®. Additionally, the study explored creating a digital twin by capturing point cloud data for specific areas.
The feasibility study revealed significant benefits: Chiba Complex alone could save tens of millions of yen annually and improve operational efficiency by over 3,000 hours. These quantified results paved the way for the company-wide adoption of Cognite Data Fusion® in April 2024.
Solution
To maximize impact, the decision was made to input data within six months and simultaneously deploy Cognite Data Fusion® across all four manufacturing sites. Ensuring data quality was a top priority. Hanako Kubo led an extensive effort to standardize approximately 10,000 piping and instrumentation diagrams (P&IDs), which had variations in format and notation across locations. Each P&ID was reviewed and reformatted in CAD to meet Cognite’s recommendations. This approach significantly enhanced the accuracy of contextualization in Cognite Data Fusion®.
Even though the project was complex, Cognite Data Fusion® allowed the team to proceed with speed, and the system was successfully rolled out at all four sites by October 2024—just six months after initiating the effort.
Results of Cognite Data Fusion®︎ implementation
Today, approximately 10,000 P&IDs and over six million data points—including repair histories, operational data, and equipment drawings—have been integrated into Cognite Data Fusion®. Employees can now instantly access this information via the system, with around 3,000 expected users in the future.
Even in its early stages, Cognite Data Fusion® has seen widespread adoption, with over 500 monthly users praising its convenient and powerful search capabilities. This positive feedback underscores the system's success in improving operational efficiency.
“We aim to go beyond merely creating a virtual integrated database. Our vision is to develop Cognite Data Fusion into a platform for advanced data utilization, enabling initiatives like automated maintenance plan optimization. By fully leveraging the system’s data model, we hope to enhance not only operational efficiency but also enable sophisticated analysis and decision-making,” said NarukiAkiyama.
“Currently, Cognite Data Fusion is primarily used as a search tool, but its potential is far greater. We are committed to driving meaningful changes in manufacturing operations through this platform,” added Hanako Kubo.
Future Prospects
Cognite Data Fusion® has revolutionized how Idemitsu Kosan collects and organizes data. By centralizing related data into a single system, maintenance planning and decision-making have become more streamlined. For instance, P&IDs, equipment records, and operational histories can now be easily consolidated, enabling quicker and more informed decisions.
Looking ahead, Idemitsu Kosan plans to further enhance its capabilities by incorporating AI-driven forecasting and decision-support tools. These advancements aim to develop systems capable of proactively identifying and mitigating risks—essentially acting as “a mechanism to notice”—to enable more sophisticated decision-making and enhanced operational safety.