Deployment of a Hybrid Machine Learning and Physics Based Drilling Advisory System at the Rig Site for ROP Optimization
Bibliographic record
Abstract
Abstract During well construction, automatic monitoring of the sensor signals for drilling dysfunction detection through pattern recognition algorithms is key to improving rate of penetration (ROP) and preventing tool failure. The addition of physics-based models can enable further improvement, but often one is limited by the contextual data needed by these models, as well as the computational power available at the edge. This paper details the successful field deployment of a system that address these challenges. The dysfunction tracking algorithms used were built using Bayesian networks as base models and validated using downhole data. Physics based models in the advisory system are used to compute the first five modes of natural frequencies for axial, torsional and lateral vibration. The contextual data required for the calculations consists of the bottom hole assembly (BHA) and survey data. Scripts were deployed to transfer this data directly from the operator's database to rig site. This system has been deployed on rigs in the US for over 4 years now, and the fact that they are being actively used to this day is a testament to its success. A key enabler here is the automatic transfer of contextual data from the office database to the rig site. The contextual data used in the model is something the crew have to input into the office database outside the needs of the advisory system. So, a process was already in place to properly record this information, and that worked to the advantage of the system. Eliminating the need to re-enter this data at the rig site was key to the success of this advisory system. Using the physics-based model, critical RPM bands are plotted on the drilling advisory screen to alert the driller whenever they are near an RPM that needs to be avoided. Visual indicators on a weight on bit (WOB)-RPM grid provide guidance to the driller on which direction to move the parameters to avoid dysfunctions and optimize drilling. Physics based models nicely complement data-based ML models in an advisory system, but real-world application of such combined systems are limited due to reasons such as timely availability of contextual data at the rig-site, or the need for contextual data that is not readily measured. In this paper, we demonstrate how the problem can be solved, and provide guidance for larger adoption of the process followed by team.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".