The drilling process is complex because unexpected situations may occur at any time. Furthermore, the drilling system is extremely long and slender, therefore prone to vibrations and often being dominated by long transient periods. Adding the fact that measurements are not well distributed along the drilling system, with the majority of real-time measurements only available at the top side and having only access to very sparse data from downhole, the drilling process is poorly observed therefore making it difficult to use standard control methods. Therefore, to achieve completely autonomous drilling operations, it is necessary to utilize a method that is capable of estimating the internal state of the drilling system from parsimonious information while being able to make decisions that will keep the operation safe but effective. A solution enabling autonomous decision-making while drilling has been developed. It relies on an optimization of the time to reach the section total depth (TD). The estimated time to reach the section TD is decomposed into the effective time spent in conducting the drilling operation and the likely time lost to solve unexpected drilling events. This optimization problem is solved by using a Markov decision process method. Several example scenarios have been run in a virtual rig environment to test the validity of the concept. It is found that the system is capable to adapt itself to various drilling conditions, as for example being aggressive when the operation runs smoothly and the estimated uncertainty of the internal states is low, but also more cautious when the downhole drilling conditions deteriorate or when observations tend to indicate more erratic behavior, which is often observed prior to a drilling event.

The drilling industry has made significant progress on physics-based torque and drag (T&D) models that can run either offline (pre-job or post-job) or in real-time. Despite its numerous benefits, real-time T&D analysis is not prevalent since it requires merging real-time and contextual data of dissimilar frequency and quality, along with repeated calibration whose results are not easily accessible to the user. Our goal is to implement a rig-based T&D advisory system which overcomes these obstacles. The first step towards real-time T&D analysis is a reliable data acquisition and processing system at the rig site. This includes the ability to receive and process data of different frequency and merge it with contextual data. Once this was accomplished, the system was implemented on more than 20 rigs in North America. We then adopted a soft-string T&D model to be used for various purposes including the automatic detection of overpull/underpull events and the depths where these occur, open-hole and casing friction factor determination, sensor calibration and real-time broomstick plotting and field data comparison for subsequent casing run design. In this paper, we demonstrate the field and office application and usage of a real-time T&D model. The system on which the model is run must be able to merge both real-time (hook load, torque, rig state, etc.) and contextual (BHA composition and specifications, wellbore design and trajectory, mud weight, etc.) data. Given the developed infrastructure, the drilling engineers have access to automated model calibration in real-time which enhance the reliability and repeatability of results and also contribute to time/cost savings. Using the embedded rig state identification engine, different real-time data points can be classified (e.g., slack-off, pick-up and rotating off-bottom) and used in T&D calibration. In addition to traditional broomstick plots, the algorithm uses probabilistic data analytics approaches to identify troublesome zones (e.g., overpull/underpull locations). In a fully automated manner, the platform generates predictions based on calibrated friction factors to enhance subsequent casing run as well. The outputs are used in both field and office in a variety of ways to improve drilling performance and safety. Using the developed platform, we automated the process of T&D analysis and reduced/eliminated the time/cost required to run physical models offline. Using data from multiple BHA runs and one casing run from an exemplary well in North America, we were able to demonstrate the benefits of the automated real-time application in comparison to the traditional offline use of torque and drag analysis.