In Houston, O'Neil and Linegar found other engineers already deep into several plans to fight the blowout.
Two dozen times they tried and failed to revive the blowout preventer, a massive apparatus of rams and valves designed to pinch off the well pipe in case of an unexpected surge of petroleum. Throughout the process, a small-scale model of the device sat on a table in one of the rooms. It seemed every time someone touched it, something fell off.
Early in May, the team moved to Option 2: the containment dome.
The dome dropped toward the seafloor for hours on the evening of May 7, as O'Neil and Linegar watched with 50-odd fellow engineers. Finally, the dome reached the spill source. Oil spilled out of the dome's door. Robot cameras showed what appeared to be shadows on the dome's underside — "until you realized," O'Neil said, "that the way the light was, shadows shouldn't be there."
When the cameras shifted, the engineers could see sooty black beehives under the dome — icy gas formations of methane that buoyed the structure and left it useless. Near 1 a.m., officials called off the mission. Engineers who had worked 20 straight hours went home, discouraged.
They returned to the command center by 6 a.m. Three hours later, the team had settled on half a dozen fresh ideas.
Because money is no object, engineers order parts as soon as they dream up a new plan. "If we build a $100,000 piece of equipment and we don't use it, it's not the end of the world," Wells said.
There's no shortage of government help, either. Customs and immigration officials have helped expedite import of parts that didn't exist in the United States — and the arrival of scientists from other countries.
Chu's team settled into a diagnostic role, using supercomputers, gamma-ray imagers and other cutting-edge tools to help BP engineers answer fundamental and vexing questions about the pressure levels in the pipe and how much force it could handle.
They helped BP build "decision trees" — "Choose Your Own Adventure" books of the scientific process, where engineers plan responses for every contingency they can imagine. In the day-to-day operation of the command center, Chu's team members are always whispering in BP's ear: Did you think of this? What will you do if it happens?
The government engineers say they're energized by the challenge. "These are the kind of problems I love," Chu said, adding later: "It's really roll-up-your-sleeves, detailed stuff."
Diagnostics aren't the only big problem for the engineers in Houston, though. There's the maddening task of managing boat traffic above the leak, so ships can stay nearly still to manage their robot workers underwater.
There's the frustration of watching deep-water robots plod through even the simplest tasks, such as tightening bolts.
"It's a different world," said team leader Tom Hunter, the director of Sandia National Laboratories, who has worked on shallow-water oil rigs and set up containment systems for underground nuclear weapons tests. "The thing that I notice mostly is the things you think would be simple, a mile beneath the surface."
But evidence of the difficulty flashes every day on video screens in the Hive: clouds of black crude billowing unabated from the pipe.
"It's like trying to do an operation on the moon," said Thomas Bickel, deputy chief engineer at Sandia and a member of Chu's team. "It's the same complexity. It's the same difficulty. And you don't have the luxury of being in an academic environment where you can work on it for three years. Everybody's very aware of that pressure."
Lately, engineers have rehearsed the "top kill," which will pump drilling fluid, or a rubbery mixture dubbed the "junk shot," or both, into the well. They have made dry runs on a blowout preventer elsewhere in Houston. In the command center, they've been "killing it on paper," Linegar said, going step by step through the process, game-planning for every possible problem. The stakes are high: Poorly executed, the top kill could blow the top of the blowout preventer and dramatically increase the oil spill's volume.
If there's irony in a company and a government taking such pains to avoid missteps — after not having a detailed response plan in the first place — the engineers have no time to focus on it.
They're so busy, in fact, that hardly anyone gathered in the Hive one night last weekend as the team notched its biggest success, inserting a catheter-like tube into the leak and piping some of the oil to a holding ship on the surface.
When engineers reported at 6 a.m. the next day, there were no big celebrations.
They still had a leak to plug.