Improve the crude oil separation process
Many oil and gas companies face the challenge of their crude oil fields struggling to meet oil quality export requirements due to too-high water content.
As a part of its efforts to continuously optimize production across its assets, a national oil company wanted to tackle this challenge.
Machine learning could help the operator predict oil quality and optimize the separation process, but the development process was costly and time-consuming.
The challenges included:
- Solution architects needed to set up an on-premises compute service, a time series database, and a visualization service.
- Data owners needed to be convinced that the data would be handled securely.
- The security department needed to explore multiple security and authentication solutions to find one that its stakeholders would approve.
- Data needed to be manually extracted from the PI System, tagged, and repaired.
- There was no way for the company to scale the solution, or to monitor when the machine learning model didn't run or the data source was down.
A data foundation powered by Industrial DataOps
Cognite delivered a live, physics-guided machine learning model to help the national oil company identify factors causing poor oil separation with recommendations for how to improve separation.
To build the solution for one oil train, more than 350 sensors were evaluated. More than 100 sensors were used in the live solution.
The model runs on top of the Industrial DataOps platform Cognite Data Fusion®, which solves the data challenges that the company faced.
Cognite Data Fusion® features:
- Built-in features and integrations with market-leading services that simplify data management.
- Security requirements that align with industry standards, secure software development life cycle, partnerships with cloud service providers, and resilience measures.
- AI-powered contextualization services that transform and find connections between different kinds of data.
- A wide selection of open-source and custom data extractors.
- Tools for model hosting, data quality monitoring, solution scaling, and more.
Operational efficiency at scale
By taking the foundational approach to solution deployment with Cognite Data Fusion®, the national oil company now has the ability to quickly scale solutions across its entire fleet of assets.
The initial deployment of Cognite Data Fusion® for just one oil train at a large separation facility delivered a time reduction of 70%, resulting in gains of more than $11.5 million by improving the quality of separated oil.
The second phase—scaling to four additional oil trains—brings potential gains of more than $75 million.
Finally, the third phase will scale Cognite Data Fusion® to the full fleet of separation facilities. This could lead to additional potential gains of more than $500 million.