Industries / Energy, water and waste / Oil and gas extraction / Offshore drilling
Seabed equipment
Seabed equipment is the hardware that sits on the sea floor or hangs below the rig during drilling: the wellhead that anchors the well, the riser pipe that links it to the rig, the valve trees that later control the flow, and the remotely operated vehicles that work where divers cannot. The blowout preventer, which also sits on the sea floor, is covered under well control.
5
Milestones achieved, from 1962 to 2018.
1
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1962
Wellheads move to the seabed
When Blue Water 1 began its record well in 297 ft (91 m) of water for Shell in January 1962, divers could work only to about 150 ft. Shell's engineers built a system to set the wellhead, the blowout preventer and the frame that guides them into line on the sea floor by remote control, and fitted the riser with the first telescoping joint, a sliding section that lets the rig rise and fall on the waves; the joint became standard in floating drilling. Shell taught its methods to other companies at a three-week school in 1963, according to historian Tyler Priest. Offshore magazine's industry timeline dates Shell's first subsea well completion in the Gulf of Mexico to 1961.
Credit
Shell Oil's Bellaire Research Center (Ron Geer, Lloyd Otteman, John Lacey, Dee DeVries)
1962
A robot works the wellhead
Howard Shatto and other Shell engineers on the Pacific coast built an underwater completion system that needed no divers. Its "Mobot", a free-swimming robot designed by Hughes Tool, was driven by propellers, guided by sonar and a television camera, and rode a circular track around the wellhead while its mechanical arm turned lock screws, worked valves and attached control hoses and guide lines. Offshore magazine's timeline lists it in 1962 as the "robotic forerunner of ROV", the remotely operated vehicles now used on deepwater wells.
Credit
Howard Shatto and Shell Oil's Pacific Coast engineers; Hughes Tool
2003
A riser parts at depth
In 2003 a drillship working in about 6,000 ft of water lost its riser, the large pipe that links the rig to the well, while pulling drill pipe from the hole. The riser split between joints 39 and 40, about 3,200 ft down, and spilled the synthetic drilling mud inside it; most of the rest lay scattered on the sea floor around the wellhead. The blowout preventer's automatic "deadman" system closed the well and cut the drill pipe, and an ROV then plugged a hydraulic connector called a hot stab into the stack to close a second set of rams. The US Minerals Management Service reported it without naming the operator or rig.
Lessona riser can fail with no warning, so the equipment on the sea floor must be able to shut the well with no help from the rig.
Credit
US Minerals Management Service (report)
2008
Subsea trees go electric
A subsea tree is the set of valves placed on a finished well to control its flow, and conventional trees are worked by hydraulic fluid piped from the surface. In September 2008 gas began to flow through an all-electric subsea production system at Total's K5F field in the Dutch North Sea, in 40 m (131 ft) of water and controlled from a platform 18 km away. Its valves and chokes were moved by electric motors, springs closed the valves if power or communication was lost, and no hydraulic fluid was released into the sea. Offshore magazine called it the world's first use of such a system.
Credit
Cameron (CameronDC system); Total E&P Nederland
2018
Heavy stacks tire wellheads
Rigs built for deep water carry much bigger blowout preventers, and when such a rig drills in shallower water the extra weight and motion bend the wellhead and conductor, the steel pipe that holds the top of the well in the sea floor, back and forth. For a well in 300 m of water, engineers at MCS Kenny calculated a minimum conductor fatigue life of 14 days with a sixth-generation rig attached, against 168 days with a third-generation rig, and found the preventer's size and weight to be the largest cause. Because fatigue analyses by different firms gave "diverging results", DNV GL and 16 oil companies wrote a common method, published in 2018 as recommended practice DNVGL-RP-E104.
Lessona rig chosen for its deepwater strength can wear out a shallow-water well faster than an older, lighter one.
Credit
John F. Greene and Dara Williams (MCS Kenny); DNV GL and its joint industry project partners
Watch for
Drones that live underwater
Saipem says its Hydrone-R, an underwater drone that stays on the sea floor between jobs instead of being launched from a ship each time, has logged more than 500 days on the seabed at Equinor's Njord field off Norway, including 240 days in a row, and has run hundreds of missions with no support vessel. It says that at the end of 2025 the drone supported commissioning work on a Njord well in waves up to 12.5 m high, when vessel-launched work-class ROVs could not operate.
Achieved whenEquinor or another outside party publishes the drone's availability and the share of subsea jobs it took over from vessel-launched ROVs, or a second operator puts resident drones into routine service on its wells.
Claimed by
Saipem, February 2026
Open
01
An open wellhead fatigue model
Free, open-source software that works out how fast waves, acting through the rig, riser and blowout preventer, use up the fatigue life of a subsea wellhead and conductor, following the steps of DNV's published method, with worked test cases anyone can rerun.
The briefCredit · unclaimed02
Drilling ROV tasks in simulation
A free simulated test of the jobs an ROV does on a drilling wellhead and blowout preventer: pushing a hot stab (a hydraulic plug) into its socket to close a ram, turning a valve with a torque tool, and holding still against a panel while doing it, with baseline robot policies trained from a few recorded demonstrations.
The briefCredit · unclaimed
Open 01 · the brief
An open wellhead fatigue model
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
Free, open-source software that works out how fast waves, acting through the rig, riser and blowout preventer, use up the fatigue life of a subsea wellhead and conductor, following the steps of DNV's published method, with worked test cases anyone can rerun.
Why it is still open
DNV GL wrote its 2018 method because "wellhead fatigue analysis conducted by different industry players has provided diverging results," and the software DNV built on that method, its Wellhead Fatigue Manager, is a commercial product. Open fatigue libraries such as fatpack count stress cycles but do not model a wellhead, and searches of the web and GitHub for this file found no open wellhead model.
Built with what exists
Python; OpenSees, an open finite-element program, to model the riser, preventer, wellhead and conductor as a string of beams; Capytaine, an open solver for wave loads on floating hulls, for the rig's motion; fatpack for cycle counting; the NORA3 wave hindcast from the Norwegian Meteorological Institute for sea states; recommended practice DNV-RP-E104 for the method; an AI coding assistant to draft code and tests against the method's equations.
First step
Build the third-generation case from the 2012 MCS Kenny article as a beam model and check that its first natural period moves from about 6 seconds at 100 m to about 25 seconds at 500 m, as the article reports.
Done looks like
The model reproduces MCS Kenny's published 300 m comparison (a minimum conductor fatigue life of about 168 days with a third-generation rig and 14 days with a sixth-generation rig) to within a factor of two, and a second group, using only the published code and inputs, gets the same numbers to within 10%.
Open 02 · the brief
Drilling ROV tasks in simulation
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A free simulated test of the jobs an ROV does on a drilling wellhead and blowout preventer: pushing a hot stab (a hydraulic plug) into its socket to close a ram, turning a valve with a torque tool, and holding still against a panel while doing it, with baseline robot policies trained from a few recorded demonstrations.
Why it is still open
WasserMan, posted in October 2026 and described by its authors as the first multi-task simulation benchmark for learning to work underwater mechanisms, has six tasks (press a button, rotate a valve, open a hatch, collect a shell, push a slider, pull a lever) and none that inserts a hot stab or a torque tool. For the European DexROV project, Ioannis Havoutis and Sylvain Calinon showed in 2017 that a robot arm could learn a hot-stabbing motion from a few demonstrations, but that work is a learning method, not a shared test with scored tasks. The job is real: after the 2003 riser failure, an ROV's hot stab closed a second set of rams on the well.
Built with what exists
WasserMan's released code and its reference ACT and diffusion policies; Isaac Lab, NVIDIA's open robot simulation framework, on which WasserMan runs; NVIDIA's GR00T 2 robot model, which NVIDIA says reaches 95% success from five demonstrations per task (see the Note "Five demonstrations per task"); simplified socket and tool shapes modeled on the ROV interface standard API 17H.
First step
Add one hot-stab socket to a WasserMan scene and get a scripted expert to complete the insertion.
Done looks like
Three tasks, each run 100 times with randomized current, visibility and actuator delay, with success rate and peak contact force published for at least two policies, and the code and scene files released under an open license.