AI, Influencers, and Grit: How Apache’s 8-Year Quest To Build Its Own Drilling Advisor Achieved Full Adoption
Bibliographic record
Abstract
When oil prices tumble, upstream research and development projects are among the first casualties. Many are put on the shelf. Few are ever taken off. Then there’s what happened at Apache Corp. More than a decade ago, the Houston-based independent oil and gas producer, which following a restructuring is now a subsidiary of APA Corp., was among a small cadre of operators working at the fore of automated drilling technology. As a first adopter, Apache had gone so far as to develop a newbuild automated rig design for its onshore shale fields in Texas. Then in mid-2014, crude prices plummeted nearly 60% over a 7-month period. The heady days of $100/bbl oil were replaced by the era of “lower for longer,” derailing Apache’s ambitious plan to build its pioneering prototype. That could have been the end of the story. But less than a year after the ambitious capital project was scrapped, something else emerged in its place. While AI-based automation was out, Apache’s drilling team was given a chance to develop the next best thing: an AI-based drilling advisory system. “We saw that there was at least a small opportunity to do something more with our rig data—that 1-Hz real-time data—by combining it and mashing it up with contextual data so that it could be something useful,” said Michael Behounek, a former director of drilling, completions, and workover performance and leader of the project for Apache where he spent the past 13 years before taking early retirement. Now a managing partner of an upstream digital consulting startup called Emerja, Behounek spoke at the recent SPE Annual Technical Conference and Exhibition while presenting SPE 215132. The paper outlines how after 8 years that small opportunity turned into a 10% year-over-year reduction in drilling costs. The system, a presumed multi-million dollar value creator, was adopted across all of Apache’s contracted rigs in 2018, playing an increasingly important role in the drilling of more than 1,700 wells across a wide spectrum of geologies. This track record spans nine onshore and offshore basins, with deployments of the system in the Permian Basin, Egypt, Canada, offshore Suriname, and the North Sea. Behounek said most of the reported cost savings stem from the models’ ability to trim rig time by predicting problems that would keep drillers from staying on bottom and turning to the right. That said, the paper, which is coauthored by Apache’s software partner Intellicess Inc., emphasizes that “the system only enables the opportunity—it is the field personnel and engineers taking the proper actions and decisions offered by the system that deliver the improvement.” Apache has shared several papers about the various components of the advisory program over the years but the most recent offers a holistic view of the strategy that led to companywide adoption. Of the dozens of takeaways it offers, some of the biggest follow.
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 | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.001 |
| 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".