Geosteering is a sequential decision-making process with the objective to maximize revenues from hydrocarbon production whilst minimizing costs. Stratigraphy-Based Steering (SBS) is a popular approach to steering horizontal wells in layered geology. SBS software helps a geosteerer to interpret the logs generated and update a geomodel around a well and extrapolate that geomodel's trends in support of well placement decisions. Such geosteering is a subjective process with a lack of data-based, quantitative universal guidelines for making geosteering decisions based on information provided by SBS tools. Since every well is different, it is challenging to find best practices without conducting controlled experiments and comparing strategies of different experts.
This paper analyzes data from 349 geosteerers, who participated in the Rogii Geosteering World Cup (GWC) 2021. The data includes interpretations and decisions for two wells representing conventional and unconventional drilling operations. The true geomodels used to generate the logs for the participants' wells were based on real fields with various formation dips and faults. After every drilled segment, the geosteerers interpreted the new data from that segment and decided where the drill bit should target to maximize their scores. These scores were calculated based on in-reservoir contact, reservoir quality, and drilling speed. This process was repeated every 2 minutes with all the user-generated information recorded for further analysis. Snapshots of interpretations of the log data for each participant's well and their corresponding decisions, recorded after every step formed a database of over 10,000 snapshots.
We conduct a statistical analysis of the recorded data to investigate the causation and correlation in the recorded data. The objectives of this paper are threefold: (1) improve our understanding of the factors leading to good geosteering decisions and well placements, (2) evaluate the extent to which good well placements are the result of interpretation and decision skills versus luck, and (3) contribute to the development of universal guidelines for making geosteering decisions based on the information from SBS tools.