This paper aims to demonstrate the application of a new automatic geosteering method that combines probabilistic interpretation with artificial intelligence (AI) for look-ahead decision-making. We expand on our previous synthetic workflow by testing our geosteering robot, named “PluRaListic,” into a synthetic environment modeled after a commercial cloud-based geosteering environment from the ROGII Geosteering World Cup (GWC). This synthetic setup allows for comprehensive testing and validation of the robot’s capabilities in a controlled yet realistic setting, ensuring that the methodologies developed can be robustly assessed before potential deployment in real-world operations.
Our automatic geosteering method combines a reinforcement learning (RL) algorithm with the particle filter (PF) method. PF continuously assimilates real-time log measurements obtained during geosteering operations, producing hundreds of most-likely geology interpretations. Simultaneously, RL uses the information gathered from PF outputs to optimize steering decisions.
The robot implementation automatically collects the new well trajectory and logs and passes the latest data through the PF. The RL uses the most-likely interpretations to balance the short- and long-term steering priorities and outputs a single recommendation that the robot sends back to the synthetic environment.
The operation of our robot significantly surpasses real-time operation requirements, making one steering decision in approximately 4 seconds, far below the 2-minute-per-stand drilling time allocated for the GWC. After running 1,000 simulations, the median outcome achieved 77.3% reservoir contact, placing the robot in the top 14% of human experts. Moreover, the robot’s best attempt surpassed all experts, highlighting its potential to exceed human expertise in optimal scenarios.
This work represents a radical innovation in geosteering that contributes to the advancement of automated geosteering frameworks. Future developments will focus on improving the performance and reliability of our robot. We also aim to enable an interactive framework for seamless collaboration between human experts and the robot, combining human expertise and consistent AI decision-making to achieve more precise and efficient drilling operations.
This abstract is taken from paper SPE 218444 by Ressi B. Muhammad and Yasaman Cheraghi, University of Stavanger; Sergey Alyaev, NORCE; Apoorv Srivastava, Stanford University; and Reidar B. Bratvold, University of Stavanger. The paper has been peer reviewed and is available as Open Access in SPE Journal on OnePetro.