Gulf Spill One Year Later: Lessons for Robotics

20 April 2011—A year ago today, an explosion onboard the Deepwater Horizon oil-drilling rig in the Gulf of Mexico killed 11 workers and sank the rig, starting the largest oil spill in American history. Government scientists estimate that a total of 780 000 cubic meters of oil seeped into the sea before BP successfully cemented the well in September.

The spill could have been much worse had it not been for the help of sophisticated remotely operated vehicles (ROVs)—car-size multiton robots engineered to be the hands and eyes of subsea operations too deep and too risky for human divers. As the world watched live streaming video of the robots working to repair the damaged rig, viewers came to understand that such robots "aren’t just handy but essential," says Tyler Schilling, president and CEO of Schilling Robotics, based in Davis, Calif., which manufactured four of the ROVs that operated at Deepwater Horizon and all of the robots’ manipulator arms.

The Gulf spill was the biggest test yet of underwater robotic technology, but it has left robotics experts divided over what technological innovations, if any, are needed for the next generation of ROVs. While some industry experts argue that ROVs are already as sophisticated as their pilots need them to be, others say that ROV technology, though dependable, is less than ideal. They predict that as offshore drilling operations get larger and go deeper, and as new government regulations demand more maintenance procedures and safety tests, rig operators will need ROVs to perform more tasks in a more reliable way.

Rig operators typically keep one or two robots on hand to make routine visual inspections, and at times, they enlist as many as six ROVs to test and assemble equipment underwater. To help contain its gusher at 1500 meters below the sea surface, BP deployed an unprecedented 16 ROVs, each piloted from shipboard control rooms and operated by a combination electrical-optical cable. BP claims that more than 4 million people watched the ROVs’ live camera feeds as the robots used their manipulator arms to try to activate the failed blowout preventer, saw off the busted well pipe, and install a temporary oil-collecting cap over the well, among other tasks. In the end, their efforts could not permanently seal the spewing well. That would require the drilling of relief wells to intersect the damaged well 4 kilometers below the seafloor and pump in cement. But experts don’t blame the ROVs and say that better robot technology wouldn’t have prevented the disaster nor ended it sooner.

"ROVs were not the limiting factor in all the work that went on in trying to stop the leak," says Drew Michel, president elect of the Maryland-based Marine Technology Society and has worked in the ROV industry for 44 years. "The capacity of ROVs has been beyond the need for the last 20 years. They are a tool looking for a problem, and [the explosion at Deepwater Horizon] came up as a problem."

Still, some engineers believe that ROV technology will need new innovations as drilling operations expand. Of course, offshore operations have all but shut down since the U.S. government suspended deepwater drilling in the Gulf last July. And although the moratorium was lifted in October, activity has been slow to pick up. But the current dry spell is just a "hiccup," and operations will soon grow bigger and venture into deeper waters, says Julio Guerrero, founder and chief scientist of Cambridge Research and Technology, an oil-industry consulting company in Massachusetts. Rigs will have to follow new government safety rules, which may demand more from ROVs, he says: "Any regulation that asks oil-drilling or oil-exploration companies to perform more careful and more meticulous activities may require more sophisticated and more timely performances from underwater vehicles."

Schilling agrees that deeper, bigger, better-regulated operations will require that ROVs do more tasks faster. But the problem, he says, is not that today’s ROVs perform inadequately but that the cost of operating them will increase. He is already seeing this happen. A few of his customers tell him that operating just one ROV control vessel, which typically supports two ROVs, costs them about US $9 per second. "As you go into deeper and deeper water, the kind of machinery you need gets larger, more complicated, and involves more ROV support activities," which pushes up the price, Schilling explains. "Over the last 10 years, I would say the cost of using ROVs has gone up by a factor of five."

Schilling’s solution is to engineer software programs to automate routine tasks that ROV pilots do now. Automation will reduce pilot errors and save rig operators time and money, he says. So far, he has made progress with two products and hopes to have them ready for clients by the end of the year.

The first piece of software helps an ROV do what Schilling calls "midwater station keeping." It lets the ROV automatically position itself in the water at any altitude above the seafloor, using an ultrashort baseline positioning system. That system transmits acoustic pulses between a transponder on the ROV and a transceiver mounted on equipment or on the seafloor. "Think of it like GPS underwater, but instead of working on radio frequencies, it’s done acoustically," Schilling explains.

The second piece of software automates routine "peg in hole" tasks using visual servoing, in which an ROV uses video input from its cameras to control its manipulator arms. Such automation would be particularly useful, Schilling says, for plugging into "hot stabs"—torpedo-shaped fluid connectors installed on seafloor equipment that let an ROV hook up to and control hydraulic systems.

New government regulations for offshore drilling require rig operators to frequently test hot stabs installed on the blowout preventer. More testing, Schilling argues, means more opportunity for error and potentially more time and money lost. A skilled ROV pilot might connect to a hot stab in 30 seconds, he says. But under bad visibility conditions, even a good pilot might miss the connection and "damage the hot stab or the receptacle, delaying the operation for hours or even days." With automating software, an ROV will take about a minute to hook up to a hot stab, he says, "but it does it successfully every time."

Michel argues that while such automating tools could be time-saving solutions, they’re not in high demand. "Any critical operation is going to have a best pilot" who can perform challenging tasks cost-effectively, he says. "It’s a difference between a want and a need. I want to drive a Ferrari, but I do just fine driving a Chevy." Many ROV pilots, he adds, also want force feedback and tactical sensors on the robots’ hands, but those features are luxuries, not necessities, and off-shore construction companies aren’t willing to pay extra for them.

Schilling, however, is convinced that new automating tools will improve accuracy and "make things happen in fewer elapsed hours or days"—attractive incentives for oil companies wanting to save money and needing to comply with government regulations. He’s not aiming to design the Ferrari of ROVs, he says; he envisions something more akin to a really good Toyota Prius. In fact, Schilling is regularly inspired by the automating innovations the automobile industry adopts: radar-based cruise control, lane-departure alerts, automatic parking. As ROVs start playing a bigger role in congested deepwater operations, he says, some of those tricks could really come in handy.