Session: 11-04-01 Drilling Automation

Submission Number: 157253

Simulator Based Testing and Validation of a Drilling Optimization System 

Rig automation is becoming common practice, due to increased instrumentation on the drill-floor and the possibility to communicate and control the drilling machines via an Automated Drilling Control System. Active control of the drilling process itself (i.e. hole creation) is an active field of innovation, with several solutions being currently developed. We consider here the testing and validation of such a drilling optimization system. This system continuously computes set-points for the three main rig machines (hoisting, rotation and circulation systems) so that the performances are optimized (highest Rate of Penetration (ROP) as possible) while ensuring that the process as a whole is still conducted in a safe manner. This involves limiting the possible range of setpoints to avoid cuttings transport issues, buckling or high downhole pressures. In this context, one of the key features is the ability to react quickly to changes in downhole conditions: abrupt formation changes and founder point excess are the two of the main scenarios that are considered. When such changes occur, it is important to react in a timely fashion to avoid unwanted situations (for example high Weight on Bit (WOB) when entering a hard formation while in ROP mode). This feature requires a continuously updated response model of the drill-bit: this relies on constant calibration of a bit-response model, that relates the surface parameters (Block velocity, topdrive RPM and pump rates) to the well's response, WOB, bit torque and estimated ROP. The system involved in this study performs this calibration using particle filter algorithms, providing constant estimates of the current formation properties and bit characteristics. The calibrated model is thereafter used to generate optimum setpoints to the machines.

Testing such a system can be challenging because of the inter-dependency between the bit-response calibration and the setpoints computation. Offline or passive testing provides little value because the calibration relies on sufficient excitation of the system: standard drilling data sets usually use constant setpoints while on-bottom, and the calibration becomes suboptimal. In turn, the setpoint generation process cannot determine optimum values due to the lack of properly calibrated models. A solution is to base the testing and validation on a simulated environment. This requires a high-fidelity drilling simulator that can be integrated with the automation system.

We present in this paper how such a simulator was used to validate the drilling optimization system under a variety of possible scenarios, which would have been impossible in a passive mode on a drilling rig. Along with the requirements on the simulation environment, the testing methodology will be presented, together with the different metrics used to validate the system.

Presenting Author: Benoit Daireaux NORCE

Presenting Author Biography: Benoît Daireaux holds a Ph. D. in Computer Science from the University of Caen, France. He has been working with drilling automation and modelling over the last 15 years, and has been part of the research and development of several automation solution now actively used in drilling operations.