Why the Deepwater Horizon blew out
On the evening of April 20, 2010, the crew of the Deepwater Horizon ran a test to prove BP’s Macondo well was sealed before they left it. The pressure said it wasn’t. The test was accepted anyway, at 8 p.m., and the crew began replacing the heavy mud in the well with seawater. At about 9:49 p.m., gas from the well ignited on the rig. Eleven of the 126 people aboard died, and oil flowed into the Gulf of Mexico for 87 days, until the well was capped on July 15. National Commission, printed pp. 1, 17, 108 to 109, 114 and 129.
The presidential commission that investigated the disaster found that three things could have held the well’s pressure back: the cement at the bottom of the well, the mud in the well and the riser, and the blowout preventer on the seafloor. Commission, printed p. 115. The test was meant to check the cement, and accepting it led the crew to take out the mud. When the blowout preventer’s blades closed, the U.S. Chemical Safety Board concluded, part of the drill pipe was outside their cutting edges, so they cut only part of it and the well kept flowing. CSB Volume 2, pp. 41 to 44.
Watch the episode
The film follows the evening through the investigators’ records, from the first pressure reading to the blowout preventer’s partial cut, rebuilt in 3D.
The test that showed a leak
The rig was finishing the well for BP about 40 miles off Louisiana, in water about a mile deep. BP counted it among its best rigs for safety: in seven years of drilling, it had not had a single “lost-time incident.” That afternoon, four executives from BP and Transocean, the company that owned the rig, had flown out to tour it. Commission, printed pp. 1, 5 and 6.
At the bottom of the well, cement sealed off the oil and gas in the rock. While the well was full of heavy drilling mud, the mud’s weight helped hold them back too. Before the rig could leave, that mud would be swapped for lighter seawater, and the cement would be on its own. The negative pressure test was the only check that the cement would hold. The crew takes the pressure off the drill pipe and watches: if the well is sealed, nothing gets in and the pressure stays at zero. Commission, printed pp. 105 to 107; CSB Volume 1, p. 27.
Three times that evening, the crew bled the pressure off the drill pipe. Three times it climbed back, the last time to 1,400 psi. During the first try, the touring executives came through the drill shack, and at least one of them could see the crew was having “a little bit of a problem.” The tour moved on. Commission, printed p. 107.
How it was called a pass
On the rig floor, the crew and BP’s two well site leaders talked the readings over. One explanation was a “bladder effect”: heavy mud still sitting in the riser, the wide pipe up to the rig, pressing on a seal in the blowout preventer, with that pressure showing up in the drill pipe. The commission later found that the 1,400 psi “could only have been caused by a leak into the well.” Commission, printed pp. 107 to 109.
After a long discussion, they ran the test again on the kill line, a pipe about 3 inches across that ran from the rig down to the same well. Both lines led to the same place, so both should have read the same. The kill line read zero, and nothing flowed from it for half an hour. The drill pipe still read 1,400. Commission, printed pp. 107 to 108.
The investigators thought the kill line may have been blocked by the spacer, the fluid pumped down between the mud and the seawater. BP had it made from two thick leftover fluids so they would not have to be shipped ashore as hazardous waste, and no one on the rig had used anything like it that way before. The CSB said a fluid that thick could have plugged a line. Commission, printed pp. 106 to 107; CSB Volume 1, pp. 27 to 28.
At 8 p.m., BP’s well site leaders, in consultation with the crew, accepted the kill line’s zero as proof that the well was sealed. The drill pipe’s 1,400 psi was never explained. Neither BP nor Transocean had given the crew a written procedure saying what a passing test had to show. Commission, printed pp. 2 and 108 to 109; CSB Volume 1, pp. 27 to 28.
An hour of warning signs
With the test called a pass, the crew began replacing the rest of the mud with seawater at 8:02 p.m. The leak the drill pipe had shown was still there. For about an hour, nothing seemed wrong. Commission, printed pp. 108 to 110.
At about 9:01, drill-pipe pressure began to rise, about 100 psi over seven minutes, while the pumps pushed seawater in at the same rate. A rise like that can mean oil or gas is entering the well and pushing the mud up, which crews call a kick. Kicks have to be caught early because gas expands as it rises: at this well, a barrel of gas could grow more than a hundredfold on the way up from the seafloor. Pressure alone doesn’t prove a kick, but an unexplained change calls for a check: stop the pumps and see whether the well keeps flowing. Commission, printed pp. 109 to 111.
At 9:08 the crew did stop the pumps, to check the spacer for oil before sending it overboard. For six minutes nothing was pumped in, and the pressure should have held level or fallen. It rose about 250 psi. The commission later called that “a clear signal that something was happening in the well.” No one appears to have noticed, and when pumping resumed the signal was hidden. The returns were now going straight over the side, which bypassed the mud tanks and a flow meter, the tools for spotting the clearest sign of a kick: more fluid coming out than going in. Commission, printed pp. 109, 112 and 121.
Shortly before 9:30, the driller noticed the drill pipe and kill line showing different pressures again, echoing the test. The crew stopped the pumps, and over the next five and a half minutes the drill-pipe pressure rose 550 psi. They bled it off, and it climbed straight back. The commission said this should have led the crew to check for flow and close in the well. Neither happened, and investigators couldn’t fully explain why. They pointed to the other work underway and a monitoring system that depended on the right person seeing the right data at the right time. Commission, printed pp. 112 to 113 and 120 to 121.
The blowout
At 9:39 the pressure began to fall, which the commission called, in retrospect, “a very bad sign”: oil and gas were likely pushing the heavy mud up past the drill pipe. Within minutes, mud was spewing from the rig floor, which appears to be when the crew first realized they had a kick. They sent the flow to the mud-gas separator and closed a seal around the drill pipe in the blowout preventer. Commission, printed pp. 113 to 114.
At about 9:45, a crew member called a supervisor: the well was blowing out, and mud was reaching the top of the derrick. Gas was already above the blowout preventer, expanding as it rushed up the riser. A Transocean representative later compared it to “a 550-ton freight train hitting the rig floor.” The separator, built to take gas out of small amounts of mud, was overwhelmed, and at about 9:49 the gas ignited. Commission, printed p. 114.
Why the blowout preventer didn’t seal the well
On the seafloor, the blowout preventer had a blind shear ram: a pair of blades built to cut the drill pipe and seal the well. After the explosion, a supervisor on the bridge pressed the emergency disconnect button, which should have closed the blades and released the rig from the well. There was no sign it worked. CSB Volume 2, pp. 19 to 21 and 30; Commission, printed pp. 114 to 115.
Another backup, the automatic deadman, was meant to close the blades if the explosion cut the control lines to the rig, as it likely did. Two control pods could trigger it, each a backup for the other. One pod’s wiring fault had drained its battery. The other had a miswired valve that should have stopped it too, but a drained battery happened to leave that valve able to open. The CSB concluded that the deadman most likely activated and the blades began to close. At the time, neither regulations nor recommended industry practice required the deadman to be tested. CSB Volume 2, pp. 14 to 16, 22 to 23, 33 to 34 and 45 to 46. (The commission, writing in 2011 before the CSB’s tests, thought the deadman had failed.)
By then, the CSB found, the pipe had likely moved partly out of the blades’ reach. The first seal the crew closed hadn’t held, but minutes later they most likely closed a pipe ram, which did seal and stopped the flow for a few minutes. That trapped the well’s pressure below the ram and inside the drill pipe running through it, while the pressure outside the pipe above the ram kept falling as the riser emptied onto the rig. No pipe is perfectly straight, and much higher pressure inside than outside can make a slight bend grow. The board found this likely buckled the pipe away from the center of the preventer. CSB Volume 2, pp. 29 to 30 and 44.
When the blades closed, part of the pipe was outside their cutting edges. They cut through only part of it, and the steel caught between the blocks kept the ram from closing fully. The partial cut punctured the pipe, and oil and gas began flowing again. The survivors abandoned the burning rig, and the supply vessel Damon Bankston took them aboard. Two days later, the Deepwater Horizon sank. CSB Volume 2, pp. 31, 34, 41 and 43; Commission, printed pp. 17 to 18.
What changed in 2016
In April 2016, a new federal well control rule addressed these failures in U.S. federal waters. Negative pressure tests needed written procedures and criteria for a successful test, submitted for approval in advance, and any sign of a failed test, “such as, but not limited to, pressure buildup or observed flow,” had to be investigated immediately. The deadman had to be tested on the seafloor to show it could close the shear rams. Over the following years, blowout preventers also had to carry two shear rams able to cut the drill pipe, at least one of which could seal the well, and a way to keep the whole pipe within the blades’ reach. 81 FR 25888, pp. 25888 and 26024 to 26034.
The well was capped on July 15, 87 days after the explosion. Eleven people lost their lives aboard the Deepwater Horizon. We remember them and the families they left behind.
For another accident where a reading was explained away, read what the Three Mile Island control room could not see. For the same company and the same investigator five years earlier, read why the BP Texas City refinery exploded.
Primary sources and imagery
National Commission on the BP Deepwater Horizon Oil Spill and Offshore Drilling, Deep Water: The Gulf Oil Disaster and the Future of Offshore Drilling, January 2011, Chapter Four (printed pp. 105 to 129; the PDF page is the printed page plus 16). Report.
U.S. Chemical Safety Board, Macondo Investigation Report, Volume 1 (pp. 27 to 28) and Volume 2 (pp. 14 to 46), June 5, 2014. Volume 1; Volume 2.
Bureau of Safety and Environmental Enforcement, Oil and Gas and Sulphur Operations in the Outer Continental Shelf: Blowout Preventer Systems and Well Control, final rule, 81 FR 25888, April 29, 2016. Federal Register.
The hero image and the preview artwork are AI-generated illustrations from the episode, not photographs or recordings of the Deepwater Horizon. The film’s 3D shots are simplified reconstructions built from the investigators’ reports and figures; the bent pipe and the blades’ partial cut show the Chemical Safety Board’s likely explanation.
