Automatic Determination of Drill-string Geometry for Numerical Drilling Models

Modern drilling automation and monitoring techniques rely heavily on the ability to predict the response of the wellbore to the drilling process. Numerical models are often used to reproduce the well’s behavior accurately and allow precise analyses of the main drilling measurements such as Stand-Pipe pressure, Hookload or Surface torque. Unfortunately, those models require a detailed configuration: drill-string characteristics, well geometry, mud properties all play a central role in the computations. In practice, the configuration process is challenging: the information may be difficult to find, be available late in the drilling process or even be erroneous. The objective of the present work is to develop statistical techniques to automatically determine the drill-string geometry from available sensor values, such that hydraulic and torque-drag models can achieve at least the same level of accuracy as if the configuration was manually entered by an operator. For an automated drill-string configuration to be of any use, it must be made available to the automation system early enough during the operations: ideally, reliable estimates should be provided during run-in-hole, and further refinements generated when circulation and rotation are first established, prior to the drilling phase. Those requirements impose to design flexible solutions that make optimum use of the few available measurements. We based our system on ensemble techniques: many real-time simulations are run in parallel to continuously maintain an estimate of the distribution of the parameters that characterize the drill-string geometry. We describe the statistical techniques developed to perform the data assimilation, as well as the necessary modifications one must apply to standard numerical models to reach a sufficient number of parallel simulations. Finally, we discuss the use of such a system in real conditions: since the drill-string geometry is a critical element of automation systems, special care must be taken when automating this part of the work procedures. We describe the different mechanisms that can be used to validate the results of the system during drilling operations. The system was intensively tested with offline data. We present and discuss the results of those tests. Of particular interest is the stability of the generated configuration as operations proceed, the quality of the configuration with respect to the predictive capabilities of the associated hydraulic and torque-drag models, and the subsequent impact on the automation system under consideration. The solution described in this contribution aims at automating parts of the configuration process associated to the use of numerical models for drilling automation systems. As per today, no such solution exists, and this process represent an important part of the manual work needed to properly run digital solutions. We therefore consider that such a system can facilitate the deployment and adoption of modern drilling automation.