Enterprise AI for Well Intervention Planning

A large vertically integrated hydrocarbon producer wants to optimize well intervention planning to achieve better field economics. Because every well requires periodic repair and maintenance, oil & gas companies engage in extensive well intervention planning to minimize production disruptions.

To optimize well intervention, upstream production engineers perform complex mathematics and physics-based calculations to model various scenarios. However, this approach presents certain challenges as the models are prone to human error, limited to certain production assets, and difficult to integrate with each other and third-party solutions. The physics-based models can also become black boxes and less accurate over time. As a result, operators struggle to synthesize various upstream production models and conduct scenario analysis to create an optimal well intervention schedule.

To enable an integrated asset modeling approach and optimize upstream well operations using AI and machine learning, the oil & gas company selected the BHC3TM AI Suite and BHC3 Production Optimization application.

Within 12 weeks, the Baker Hughes and C3 AI team created a unified, federated data image, integrated 4 mathematics and physics-based models from the oil & gas company, built a financial economic model for maximizing upstream profitability, and configured multiple optimization algorithms and the BHC3 Product Optimization dashboard.

First, the team created a unified data image by ingesting more than 3.8 million rows of data from 11 disparate data sources, leveraging the distributed processing capabilities of the BHC3 AI Suite. The data model represented data from two oil fields with over 1000 wells combined and included well reports and Vertical Lift Performance (VLP) tables, reservoir pressure- volume-temperature (PVT) properties and pressures, well production reserves and interventions, water cut and efficiency index model parameters, and financial data.

With the data model as the foundation, the team created an integrated asset model by integrating multiple reservoir, well, and infrastructure models from the company and a financial economic model built by the BHC3 team, using BHC3 AI Suite’s low-code approach and pre-integrated Jupyter environment. Then, the team configured a new optimizer that uses a Bayesian modeling approach to run on top of the integrated asset models and conducted over 50 machine learning experiments across six use cases.

Finally, the team configured the BHC3 Production Optimization dashboard with pre-built UI components to display operating KPIs and optimized scenarios. Users can create new scenarios, compare scenarios and perform model tuning.