How AI-supported Root Cause Analysis (RCA) Is Transforming Asset Reliability at Aker BP - The Journey

IN SHORT
Last year, Aker BP leveraged Cognite Data Fusion™ (CDF) and AI-powered Atlas Agents to identify technology for streamlining Root Cause Analysis (RCA) processes, aiming to improve process efficiency and reduce cost. The value potential identified has been: 

• Ability to easily and rapidly find all relevant information
• Increased number of performed RCAs with the same resources
• Significant improvement in RCA process efficiency (on data-driven tasks)
• Better insights through AI-assisted multidisciplinary data analysis
• Reduce the number of reoccurring failures and thus increased MTTF

CHALLENGE
Aker BP operates six brownfield assets, which like any asset all over the world, are expected to require an increasing number of Root Cause Analyses (RCA) as they age. However, existing RCA processes can be improved, potentially leading to:

• Lengthy and cumbersome data gathering from multiple sources and systems (PI, SAP, inspection and maintenance reports on local share points)
• Limited real-time collaboration across disciplines

SOLUTION
Aker BP tested an integrated AI-supported RCA solution powered by:

• Cognite Data Fusion (CDF) for underlying Knowledge graph with contextualized real-time data
• AI Builder and Atlas AI Agents for automated insights and easy chat interface
• Canvas and Charts for visualizing multidisciplinary data required for RCA

This solution enables engineers to:
✅ Find data efficiently through common AI Agents searching through the knowledge graph
✅ Analyze RCA insights faster in a shared workspace (Canvas)
✅ Improve decision-making thanks to easy access to the data

By deploying AI agents initially in the sandbox environment, Aker BP decided to implement the solutions further in their production environment and refine the capabilities of AI Agents, ensuring seamless integration into their reliability workflows.

IMPACT
With AI supported RCAs, Aker BP has shown value potential related to:
🔍 Ability to easily and rapidly find ”all” relevant information pertaining to a problem/issue, unbiased by user prerequisites
📉 significant improvement in RCA process efficiency (on data-driven tasks)
⚡ AI supported RCA resolution, allowing to increase the number of performed RCAs
🔍 Real-time, data-driven collaboration across disciplines 
🔍 Improved insights and learnings from RCAs across all plants