Natural Language Processing Applied to Reduction of False and Missed Alarms in Kick and Lost Circulation Detection
Bibliographic record
Abstract
Abstract Kick and lost circulation events are large contributors to non-productive time. Therefore, early detection of these events is crucial. In the absence of good flow in and flow out sensors, pit volume trends offer the best possibility for influx/loss detection, but errors occur since external mud addition /removal to the pits is not monitored or sensed. The goal is to reduce false alarms caused by such mud additions and removal. Data analyzed from over 100s of wells in North America show that mud addition and removal results in certain unique pit volume gain / loss trends, and these trends are quite different from a kick, a lost circulation or a wellbore breathing event trend. Additionally, driller's input text memos into the data aggregation system (EDR) and these memos often provide information with regards to pit operations. In this paper, we introduce a method that utilizes a Bayesian network to aggregate trends detected in time-series data with events identified by natural language processing (NLP) of driller memos critical to greatly improve the accuracy and robustness of kick and lost circulation detection. The methodology was implemented in software that is currently running on rigs in North America. During the test phase, we applied it on several historical wells with lost circulation events and several historical wells with kick events. We were able to identify and quantify the losses even during connections and mud additions, where usually pit volume was increasing despite continual losses. Also, the real-time simultaneous analysis of driller memos provides context to pit volume trends and further reduce the false alarms. The algorithm is also able to take account of pit volume that was reduced due to drilling. Quantification of the losses offers more insight into what lost circulation material to use and the changes in the rate of loss while drilling. This approach was very robust in discovering kicks as well and differentiating it from mud addition and wellbore breathing events. These historical case studies will be detailed in this paper. This is the first time that patterns in mud volume addition and removal detected from time-series data have been used along with driller memos using NLP to reduce false alerts in kick and lost circulation detection. This approach is particularly useful in identifying kick and lost circulation events from pit volume data, especially when good flow in and flow out sensors are not available. The paper provides guidance on how real-time sensor data can be combined with textual data to improve the outputs from an advisory system.
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 machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.004 | 0.019 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.003 | 0.002 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
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 source (direct Gemma or distilled Codex), 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".