top of page

WillCo Drilling Intelligence - Edition 23: Drilling Automation Moves to Fleet Scale

Writer: William B. Contreras
William B. Contreras
13 hours ago
11 min read

Updated: 5 hours ago



A four-rung ladder titled The loop climbs to scale. Task, measured: Halliburton and GeoPark, 12 automated wells versus 4 offsets. Integrated well, field qualitative: Halliburton, Delaware Basin, six-well campaign. Fleet, measured on 31 rigs: Patterson-UTI, more than 50 rigs and over 24 million ft analyzed. Commercial, company-reported: SLB, more than 750,000 ft automated in Q3 per management. Reach grows up the ladder while the evidence shifts from measured results to company statements
Figure 1. The loop climbs to scale: four rungs in this edition, each tagged with the evidence behind it. Reach grows up the ladder while the evidence shifts from measured results to company statements (WillCo analysis).

🛢️ Top Story

Halliburton and GeoPark Put the Connection Under Closed-Loop Control in Colombia

Carryover: SPE-234102-MS, published October 14 in the ATCE 2026 proceedings.


Halliburton and GeoPark put the connection under closed-loop control on 12 wells in Colombia's Llanos Basin. Total connection time fell 30% against 4 conventionally drilled offsets; the authors' economic evaluation puts the average time saved at 3.7-4.5 hours per well. The authors call the connection one of the largest contributors to hidden nonproductive time (NPT), because execution depends on the operator at the controls and wellbore conditions get limited real-time validation. There were no recordable incidents across the automated dataset, in what the authors call the first published field deployment of its kind in the Llanos Basin.



The authors say the savings came not only from faster execution but from dropping the conservative practices normally used to compensate for uncertain downhole conditions, a change the fallback rules made defensible. Execution was consistent across crews and shifts, and run-to-run connection-time distributions narrowed over the campaign. Removing manual steps from the sequence reduced personnel exposure on the rig floor.


🎯 Your move: across 12 wells, 3.7-4.5 hours per well is roughly 44-54 rig hours, about two days of rig time; price that at your spread rate, then ask your contractor for connection-time distributions by crew and shift, not just the average. A narrowing distribution is where automation earns its keep (WillCo analysis).


What makes the result usable is the written design around it (WillCo analysis). Figure 2 shows the controls, the human gate and the advisory layer; two more rules complete it:

  • Degraded mode: defined fallback rules govern the system when sensing or automation degrades.

  • Authority: drilling strategy changes by lithological interval go through management of change (MOC); the paper records the decision authority and review cadence the system ran under.


A funnel titled Automation narrows where the human looks. Every connection: the sequencer runs the steps in closed loop. Every wellbore check: gating on downhole-sensor conditions is automated. Amber and red states: a human confirms before it continues. Advisory layer: multivariate anomaly consensus, with automatic gating the authors' stated next step. Recorded alongside the results: decision authority, review cadence, degraded-mode rules, and drilling strategy changes by lithological interval under MOC
Figure 2. The governed design behind the Halliburton and GeoPark connections: the sequencer runs the connection in closed loop, wellbore gating is automated, and a human confirms amber and red states. Layers per SPE-234102-MS; the funnel framing is WillCo analysis.


⚙️ Fleet-Scale Drilling Automation

Patterson-UTI Tunes Its Auto Drillers by Basin Across the Fleet: Dysfunction Time Down 90% in Laterals

Time in auto driller dysfunction fell 90% in lateral sections and more than 60% in vertical sections across over 24 million ft drilled by 31 rigs in two years, Patterson-UTI authors report in SPE-234119-MS (ATCE 2026). The basin-specific gain schedules now run on more than 50 rigs. Drillers cut manual gain changes by 50%, which the authors read as increased trust, and the tuned controller converged to setpoints twice as fast as the original.



🎯 Your move: price half a trip per well at your own trip time and spread rate, then ask your contractor for the manual gain-change rate before and after any controller change. That rate is a trust metric already in your rig data (WillCo analysis).



🧩 Integrated Well

Delaware Basin Automation Rules Changed Mid-Campaign: Who Signs Off on Yours?


Against manual operations, Halliburton's authors report less variability between crews, better adherence to planned trajectories and more efficient transitions through complex lithologies on a six-well Delaware Basin in-fill development (SPE-234023-MS, ATCE 2026). Under supervisory human oversight, the integrated platform sent steering recommendations to a rotary steerable system (RSS), scanned for collision risk against offset wells, and recommended drilling parameters. Automation acceptance criteria were adjusted after the initial wells, folding their lessons into later ones, the authors report.


🎯 Your move: on any multi-well automation campaign, decide before the first well who can change acceptance criteria between wells, how the change is recorded, and whether it goes through MOC (WillCo analysis).



🌐 Commercial Scale

SLB Puts Automated Footage on the Record: More Than 750,000 ft in the Third Quarter

On SLB's third-quarter (Q3) call on October 23, management said the company delivered more than 750,000 ft of automated drilling in the quarter. Filed results put SLB's Digital annualized recurring revenue (ARR) at $1.04 billion, up 13% year on year.



Offshore Malaysia, an SLB autonomous geosteering system made four autonomous trajectory changes with sub-20-second decision cycles in a thin, variably dipping sand. SLB, which first published the case in May and repeated it in its Q3 release, calls it Asia's first fully autonomous geosteering operation. Halliburton is taking closed-loop control offshore too: its October 20 results listed LOGIX automation and remote operations on Eni's deepwater exploration well offshore Indonesia.


🎯 Your move: footage shows how much drilling ran in automation, not what it saved. Ask for automated footage on your wells reported next to connection time, trips or time in dysfunction (WillCo analysis).



📏 The Measurement Under the Loop

Your Auto Driller May Be Holding the Wrong Weight

A closed loop is only as good as the reference entering it (WillCo analysis).


On multiple North American wells, Intellicess and NOV found the driller's weight on bit (WOB) often tracked the static estimate rather than the dynamic one, a sign that WOB may be overestimated because axial drag goes unaccounted, per SPE-233801-MS. Any auto driller holding a WOB setpoint inherits that error (WillCo analysis).


Their fix automatically identifies the hookload that corresponds to zero WOB, taking a static and a dynamic zero before and after each connection, with a torque and drag model to reduce stand-to-stand variability. Validated with surface and downhole data across vertical, curve and lateral sections, both automated estimates were more accurate and less variable than manual zeroing.


A four-panel comic titled The loop is only as good as its zero. Panel 1: the hookload is zeroed by hand and axial drag often stays out of the reference. Panel 2: the auto driller locks onto the setpoint and every indicator is green. Panel 3: the driller says that is not what I feel; the bit may carry less weight than the screen shows, and the loop holds the wrong number perfectly. Panel 4: automatic static and dynamic zeroing, before and after each connection, with a torque and drag model, resets the reference; now I trust the setpoint
Figure 3. The loop is only as good as its zero: automatic static and dynamic zeroing resets the weight-on-bit reference at every connection. Mechanism per SPE-233801-MS; the story framing is WillCo analysis.

🎯 Your move: before trusting any auto driller setpoint, ask how the WOB zero is taken and whether drag is in it (WillCo analysis).



📷 Hole Cleaning by Camera

Shaker Cameras and Circulating Time at TD

Offshore Guyana, ExxonMobil and DrillDocs report that shorter circulating time at total depth (TD) before tripping out avoided hours of invisible lost time (ILT), resulting in hundreds of thousands of dollars in rig-time savings, the SPE-233838-MS abstract states, with shaker-camera data one of the contributing improvements. On the Noble Sam Croft and Noble Don Taylor drillships, the cameras measured the relative cuttings recovery rate (RCRR) from shaker video, validated against mud logging and weight-based cuttings flow meters.


DrillDocs said on October 21 that the system is now installed on Valaris' DS-9 drillship in Angola and on the Hebron platform in Eastern Canada, its first deployments in either basin. The company expects real-time cuttings measurement as standard practice to reduce total circulating time by up to 25%; the first measured result from either new installation is the number to wait for (WillCo analysis).


🎯 Your move: ask how circulating time at TD is decided on your wells today: a fixed rule or a measured cuttings return (WillCo analysis).



📋 In Brief

  • 🛠️ Rig controls vs. the BHA: mismatch can cost more than $100 million a year in tool damage for SLB alone, its authors say (SPE-233888-MS): SLB authors say a mismatch between the rig control system (RCS) and the bottomhole assembly (BHA) can drive that tool-damage bill before any NPT is counted, and show how the RCS itself can induce downhole vibration. 🎯 Your move: ask whether your rig control system settings are matched to the BHA before each run, not after the first failure (WillCo analysis).

  • 🤖 Kuwait Oil Company (KOC) and Ghaia.ai schedule workover rigs with a multi-agent AI system (SPE-234185-MS): the authors project a 16% higher 90-day production gain and 46% less fleet-average rig idle time than manual schedules built from the same planning inputs, across about 40 rigs and 200 wells in KOC's Greater Burgan complex. Human review and approval are retained throughout, and schedule preparation fell from about ten days to minutes. 🎯 Your move: copy the human-approval step before any agent touches a rig schedule (WillCo analysis).

  • 🧭 SLB moves the survey into the steering tool (SPE-233943-MS): the authors call their new RSS, with mud-pulse telemetry and surveying in one collar, the first at-bit definitive inclination and azimuth survey package from a commercial RSS; the survey is corrected downhole before it is pulsed to surface. 🎯 Your move: put the survey-at-the-bit question to your directional provider before the next lateral (WillCo analysis).


📡 On WillCo's Radar



Why it matters: remote control over fiber from shore is the transferable piece for anyone planning unmanned seabed operations (WillCo analysis).

  • UT Austin automates the bit-damage forensics behind redesign (SPE-233870-MS): a bit-rock interaction model reconstructs time-varying cutter forces from drilling data to estimate where and when cutters are damaged. In five Utah FORGE runs using measured 1 Hz data, the predictions correlated with observed dull grades. The authors call it the first automated cutter-scale damage forensics workflow. Why it matters: the next proof point is a redesigned bit that performs better in the field (WillCo analysis).

  • University of Houston's DrillBench (SPE-234135-MS): a benchmark of 2,155 validated operational episodes from a full geothermal drilling campaign tests AI agents on multi-stage drilling decisions. An agent that revises early hypotheses when later evidence contradicts them outperformed single-pass and non-feedback baselines across strong large language models (LLMs). Why it matters: an agent that revises its hypothesis is the behavior to require before letting one advise on a live well (WillCo analysis).



A quadrant titled Scale vs. evidence, with scale of deployment from one well to fleet or commercial on the horizontal axis and evidence from projected to measured on the vertical axis. Field-tested, still local: Halliburton and GeoPark, ExxonMobil and DrillDocs, Halliburton Delaware. Measured at scale: Patterson-UTI. Early signal: SLB Malaysia, UT Austin. Scaling ahead of measurement: SLB Q3 (company-reported), KOC and Ghaia.ai (projected)
Figure 4. Scale vs. evidence: where this edition's developments sit. Anything right of center and below the midline is scaling ahead of its measurements; ask for measured results before you buy it (WillCo analysis).

Several companies told Norway's offshore regulator they had no AI that could affect safety, yet some of the same companies described AI in well planning.


They also pointed to AI in data analysis that supports operational decisions, the Norwegian Ocean Industry Authority (Havtil) found. A system that "simply" recommends can still affect safety indirectly, Havtil said on October 23, through how its output is interpreted and what gets prioritized.


Its survey of operating companies showed varying maturity in managing AI. Some have registers, roles and processes to identify, classify and monitor AI systems. Many lack a comprehensive overview of the systems being developed, procured or used, including AI built into supplier products. Havtil will use the findings in further dialogue, guidance and supervision.


The Halliburton and GeoPark paper is the rig-level counterexample: it names what the automation controls, where a human confirms, what stays advisory, what happens when sensing degrades, and who holds decision authority (WillCo analysis). Advisory systems like the one SLB says now monitors more than 20% of active international rigs are the kind of decision support whose indirect risks Norway's HSE regulations, as Havtil reads them, require companies to identify and address (WillCo analysis).



🔭 What We're Watching Next

  • Precision Drilling: on its October 28 call, CEO Carey Ford said the Alpha and EverGreen businesses are on track for their highest revenue and contribution margin ever, that more than 100 customer representatives from 12 customers have visited the Calgary remote operations center since it opened in August, and that customer talks about robotics have deepened in the U.S. and Canada. The next proof point is a number: an Alpha rig count, automated footage or a measured drilling result. Patterson-UTI's Q3 results are the other test of the fleet figures in SPE-234119-MS.

  • Halliburton + GeoPark: does automatic gating of the anomaly consensus, the authors' stated next step, reach the field, and does the offset sample grow beyond four wells?

  • DrillDocs in Angola and Canada: the first measured circulating-time result from either new installation.

  • Havtil: what form its follow-up guidance and supervision of AI takes.


✅ Practical Recommendations


A fishbone diagram titled Why a closed loop underdelivers. Measurement: hookload zeroed by hand, axial drag left out (Intellicess and NOV). Control tuning: gains not tuned to conditions, frequent manual gain changes (Patterson-UTI). Machine: rig controls fight the BHA, self-induced vibration (SLB). Method: lessons not carried forward, no degraded-mode rule (Halliburton, Delaware and Llanos). People: crew-to-crew variability, low trust and more gain changes (Halliburton and GeoPark, Patterson-UTI). Environment: lithology transitions, basin-specific conditions (Halliburton Delaware, Patterson-UTI). Effect: the loop underdelivers at scale. Fix the cause, not the setpoint
Figure 5. Why a closed loop underdelivers at scale: six families of cause, each tied to a paper in this edition that addresses it. The grouping and the cause-to-paper mapping are WillCo analysis.

For operators and drilling contractors (WillCo analysis): before you scale a loop, write down how it behaves. Start with the first two before your next contractor performance review: they need only data you already own.

  1. 📊 Track trust this month (Patterson-UTI, Halliburton + GeoPark): pull manual gain-change and override rates, and connection-time spread by crew and shift, before and after each automation change: a falling override rate is trust you can measure.

  2. 📏 Audit the WOB zero on your next well (Intellicess + NOV): check how WOB is zeroed and whether drag is in it; a hand-set zero means the controller is tuned to an estimate.

  3. 📐 Price the footage at your next vendor review (SLB): require automated footage per well reported next to connection time, trips or dysfunction time.

  4. 🚦 Name the gates before the next spud (Halliburton + GeoPark): one page per loop covering what it controls, where a human confirms, what stays advisory, and what it does when sensing degrades.

  5. 📋 Build the register this quarter (Havtil): list every automated or advisory system touching your wells, including those built into supplier products, and mark which can influence a safety-relevant decision, directly or indirectly.


A one-page register card titled Before a loop scales, put it on the register: one page per automated or advisory system touching the well. Seven fields: system, owner and version; the decision it influences, directly or indirectly; mode, monitoring, advisory or closed loop; human gate, who confirms or overrides and in which states; degraded mode, what it does when sensing or automation fails; evidence, baseline, metric, sample size and evidence class; change control, MOC owner, review cadence and who may change the acceptance criteria. A cycle runs register, classify, gate, measure, review. A loop you can describe on one page is a loop you can defend at fleet scale
Figure 6. Before a loop scales, put it on the register: seven fields every automated or advisory system should carry. WillCo analysis, informed by Havtil's October 23 AI-governance findings and the governed automation design reported in SPE-234102-MS.

A loop you can describe on one page is a loop you can defend at fleet scale.


📚 References and Sources

  1. R. Perdomo, C. A. Zapata, C. Ramírez (Halliburton Colombia); Y. J. Yubran (GeoPark Colombia). "Eliminating Hidden Waste Through Closed Loop Automated Connections: Improving Drilling Performance in Colombia's Llanos Basin," SPE-234102-MS, SPE Annual Technical Conference and Exhibition (ATCE 2026), published Oct 14, 2026. DOI: 10.2118/234102-MS. https://onepetro.org/SPEATCE/proceedings-abstract/26ATCE/26ATCE/D011S005R008/808942

  2. A. Groh, R. Fischer, Y. Xue, C. Christopher (Patterson-UTI Drilling Company). "Reducing Auto Driller Dysfunction and Tripping Risk at Scale with Reduced-Order Physical Modeling and Gain Scheduling," SPE-234119-MS, ATCE 2026. DOI: 10.2118/234119-MS. https://onepetro.org/SPEATCE/proceedings/26ATCE/26ATCE/D031S043R002/809004

  3. C. Stich, Z. Jamayev, F. Ferreira, D. Goncalves (Halliburton). "Field Deployment of an Autonomous Drilling Platform in a Multi-Well Delaware Basin Development," SPE-234023-MS, ATCE 2026. DOI: 10.2118/234023-MS. https://onepetro.org/SPEATCE/proceedings/26ATCE/26ATCE/D011S008R003/808843

  4. SLB third-quarter 2026 results, Exhibit 99 to Form 8-K, Oct 23, 2026. https://www.sec.gov/Archives/edgar/data/87347/000119312526429435/d151271dex99.htm

  5. SLB Q3 2026 earnings call, Oct 23, 2026: LSEG StreetEvents edited transcript, hosted by SLB. https://investorcenter.slb.com/static-files/b578b6d9-7008-4e3d-9317-7112b7c08493

  6. "Major operator drills Asia's first well section delivered with autonomous geosteering, setting a new benchmark" - SLB case study, May 28, 2026. https://www.slb.com/resource-library/case-study-with-navigation/di/2026/major-operator-drills-asias-first-well-section-delivered-with-autonomous-geosteering

  7. Halliburton third-quarter 2026 results, Exhibit 99.1 to Form 8-K, Oct 20, 2026. https://www.sec.gov/Archives/edgar/data/0000045012/000004501226000070/livemastererdocument.htm

  8. J. Cortez, D. Yoon, A. Ambrus, M. Behounek, P. Ashok (Intellicess); M. Taha, M. Mahmoud, M. Elghor, G. Hickin, C. Harkey (NOV). "Automated Static and Dynamic Zeroing for Improved Weight-On-Bit Accuracy: A Multi-Well Field Validation," SPE-233801-MS, ATCE 2026. DOI: 10.2118/233801-MS. https://onepetro.org/SPEATCE/proceedings/26ATCE/26ATCE/D031S043R001/808744

  9. J. Eze, C. G. Childress, K. Kulkarni (ExxonMobil Upstream Integrated Solutions); C. Holt, L. Holt, M. Bell (DrillDocs). "Computer Vision Shaker Monitoring Reduces Lost Time in Offshore Guyana," SPE-233838-MS, ATCE 2026, published Oct 14, 2026; ATCE session slot Oct 21, 2026. DOI: 10.2118/233838-MS. https://onepetro.org/SPEATCE/proceedings/26ATCE/26ATCE/D011S005R002/808794

  10. "DrillDocs Co-Authors Paper Showing the Value of Real-Time Cuttings Measurement, Expands Deployment to Angola and Canada" - DrillDocs via PR Newswire, Oct 21, 2026. https://www.prnewswire.com/news-releases/drilldocs-co-authors-paper-showing-the-value-of-real-time-cuttings-measurement-expands-deployment-to-angola-and-canada-302910988.html

  11. M. Katayama, O. Panova, A. Whitmore, A. Croux, A. Ledroz, M. ElGizawy, R. M. Lowdon (SLB). "Enabling Single Tool Steering and Surveying: Advanced Modeling and Estimation Techniques for an MWD Integrated into an RSS," SPE-233943-MS, ATCE 2026. DOI: 10.2118/233943-MS. https://onepetro.org/SPEATCE/proceedings/26ATCE/26ATCE/D011S008R002/808817

  12. R. Aguiar, L. F. Gonzalez, M. Hadi, D. Li, G. Carbone, B. Dow (SLB). "Partners in Crime - Matching the Rig Control System and Bottom Hole Assembly to Maximize Performance and Minimize Failures," SPE-233888-MS, ATCE 2026. DOI: 10.2118/233888-MS. https://onepetro.org/SPEATCE/proceedings/26ATCE/26ATCE/D031S043R003/808759

  13. S. F. AlSaffar, H. H. Shuber (Kuwait Oil Company); M. R. M. Amcha, L. Targamadze, M. N. Ibrahim (Ghaia.ai). "A Multi-Agent AI Optimization System for Improving Rig Utilization and Production Gains in the Burgan Field," SPE-234185-MS, ATCE 2026. DOI: 10.2118/234185-MS. https://onepetro.org/SPEATCE/proceedings/26ATCE/26ATCE/D031S039R005/809019

  14. "Endurance Energy Drills World's First Subsea Geothermal Well and Runs First Net-Positive Power Plant" - Endurance Energy, Oct 19, 2026. https://www.enduranceenergy.com/press/adelie

  15. Secondary coverage: "Endurance Energy drills first subsea geothermal well at Axial Seamount" - Drilling Contractor, Oct 21, 2026. https://drillingcontractor.org/endurance-energy-drills-first-subsea-geothermal-well-at-axial-seamount-80397 Also "Endurance Energy drills subsea geothermal well, demonstrates net-positive power" - World Oil, Oct 20, 2026. https://www.worldoil.com/news/2026/10/20/endurance-energy-drills-subsea-geothermal-well-demonstrates-net-positive-power/

  16. A. Safarov, Z. Burdette, I. Xu, P. Ashok, E. van Oort (The University of Texas at Austin). "Automated Drill Bit Redesign Using a Bit-Rock Interaction Model and Time Series Drilling Data," SPE-233870-MS, ATCE 2026. DOI: 10.2118/233870-MS. https://onepetro.org/SPEATCE/proceedings-abstract/26ATCE/26ATCE/D011S008R005/808752

  17. A. Shah, M. Jamal, X. Wu, X. Fu, J. Chen (University of Houston). "DrillBench: A Sequential Decision-Making Benchmark for Drilling Operations with Cross-Stage Agent Feedback," SPE-234135-MS, ATCE 2026. DOI: 10.2118/234135-MS. https://onepetro.org/SPEATCE/proceedings/26ATCE/26ATCE/D031S043R007/809000

  18. "Artificial intelligence (AI) in the petroleum industry is in need of better oversight and management" - Havtil, Oct 23, 2026. https://www.havtil.no/en/explore-technical-subjects2/technical-competence/reports-from-projects/2026/artificial-intelligence-ai-in-the-petroleum-industry-is-in-need-of-better-oversight-and-management/

  19. "Precision Drilling Corporation 2026 Third Quarter Results Conference Call and Webcast" - Precision Drilling via GlobeNewswire, Sept. 29, 2026. https://www.precisiondrilling.com/press-release/precision-drilling-corporation-2026-third-quarter-results-conference-call-and-webcast/

  20. Precision Drilling Q3 2026 earnings call, Oct 28, 2026: transcript via Investing.com. https://www.investing.com/news/transcripts/earnings-call-transcript-precision-drilling-q3-2026-beats-profit-forecasts-93CH-4980395

  21. "Halliburton, Eni achieve industry-first closed-loop rig automation in deepwater Indonesia" - World Oil, Jul 9, 2026. https://worldoil.com/news/2026/7/9/halliburton-eni-achieve-industry-first-closed-loop-rig-automation-in-deepwater-indonesia/

  22. Prior WillCo Drilling Intelligence editions and archive - https://willcodrilling.com/blog


About this report: The WillCo Drilling Intelligence Report is a weekly synthesis of verified developments in digital drilling and well construction, published by WillCo Drilling Consulting. Analysis, forecasts, and recommendations reflect WillCo's independent engineering perspective.



📖 Glossary of Acronyms



Comments

bottom of page