Fast decision making with AI
Business
Challenge
Optimize the lifespan of an oil field by maximizing output and minimizing costs.
The complexity of daily situations exceeds a Production Engineer’s ability to uncover and synthesize all risks when making high-impact decisions. This engineer must quickly identify problems, triage issues, and evaluate root causes to avoid downtime, which has an operational cost of $1 million / hour!
Environmental
Consideration
Timely execution of critical high-impact decisions to reduce catastrophic risks that affect health, safety, and the environment in aging infrastructure.
Note: This project delivered in 2017 using “Traditional” Deep Learning.
Capabilities would be even greater in 2024 with advances in foundational models and transformer architcture.
Design principal: Market research, Product offering, UX Research, UX design, AI tooling, Visual UI design,
Meet Jason, a Production Engineer
His Objective
Jason needs to make sure the oil field is performing at optimal production efficiency at any given time.
His Challenges
He needs to avoid blind-spots to solve production shortfalls and have the right data at the right time in order to make the best decisions. He must take high-confidence actions quickly to avoid catastrophic disasters.
At 7 am Jason Inherited a Hidden Problem
Jason receives an alarm that production is down 20% for Well A-22. He begins troubleshooting before his team meeting in an hour.
Jason Needs Insights—Fast!
Jason zooms into the problem area. The steam map model indicated a breakthrough problem shouldn’t happen for another two years.
Natural Language Collaboration
Jason begins investigating by formulating a hypothesis and asking Watson questions.
Watson Correlates Similar Scenarios
He selects Well F-3 as the target and Watson visualizes correlating events.
Comparing Wells
Jason uses the timeline to compare the scenarios.
Watson “understands” similar topics from different types of data sources.
Watson reveals a similar problem occurred in 2010.
The Moment of Insight!
The root problem is likely a fracture, induced 9 months ago in Well A-22.
Further investigation with Watson reveals an appropriate action to save $ millions.
Knowledge
Without decades of experience, Jason can form a hypothesis and understand the root cause in about an hour, using Watson. Jason also avoids spending days locating, waiting, and researching files, ultimately saving at least $20 million.
