Robust and Automatic Drilling Model Configuration

The goal of the project is to remove or reduce the need for manual configuration of the systems used for drilling automation. Today drilling automation systems need to be manually configured and several parameters are required for the systems to behave correctly. This need for detailed input usually reduces the usability and flexibility of automated drilling system and hinders their large scale deployment. The main configuration component correspond to the detailed description of the elements involved in the drilling operation: the fluid system or the drill-string geometry are for example complex structures mandatory for reliable simulation of the on-going drilling process. It is the main objective of this project to automate the configuration process, and therefore eliminate the need for manual inputs. As a first goal we will develop simpler models that emulates the more complex models used in automation as state of the art and at the same time design methods to properly control the induced modelling errors. Those simplified models will be combined with the available measurements to automatically infer the configuration that should be used by the drilling automation system. The final goal is to gain knowledge on where to locate sensor data and to find out which sensor data is required to make a smooth and accurate configuration of the system.

-

The goal of the RACO project is to automate the configuration of the drilling models used for drilling automation systems, as this task is nowadays performed manually prior to each operation. The numerical models involved in those systems are high fidelity transient models, that allows precise simulations of the hydraulic, mechanical and thermo-mechanical behavior of the wellbore during drilling operations. To achieve the predictive capabilities that automation requires, adequate description of the main components of the drilling system is mandatory. We propose to replace the risky and time consuming manual configuration work by an automated procedure, that will use the available sensor data to deduce the system’s main characteristics, such as the pipe dimensions and elastic properties, and the fluid’s base properties. The objectives of the project are very ambitious, due to the difficulty of the task: the numerical models are time consuming and the measurements may be sparse in space and time. We will therefore adopt a two step strategy. We will first develop simplified models: the gain in computation time will enable proper exploration of the configuration space. Then, parameter estimation schemes will be developed, tailored for sparse sensor data and proxy models. A final achievement of the project will a procedure to estimate the optimal sensor set-up prior to operations. The research done in the project will impact several aspects of drilling automation: it will considerably ease the development of advanced systems, therefore improving reliability and decreasing cost. The model simplification techniques will potentially enable new automated functions, presently unfeasible because of computational cost. The drilling data assimilation schemes will lead to improved state identification, and safer drilling operations.