Something was wrong.
BP was preparing to plug its well 50 miles off the Louisiana coast, and longtime technical adviser Jesse Gagliano was running computer models to finalize details.
"We have a potential problem here," the Halliburton employee told three colleagues he met in the hallway in BP PLC's Houston headquarters. He said his computer model was predicting a "serious gas flow problem" with BP's well-abandonment plan.
His idea for addressing the issue never would be carried out. BP decided it wasn't necessary. Five days later, on April 20, the well blew out, causing the worst offshore spill in U.S. history. In an internal report released Wednesday, BP stood by its decision, saying Gagliano's plan would not have stopped the explosion.
The disagreement was just one of several that emerged in the days and hours before the blast, according to BP's report and e-mails, documents and testimony gathered by federal investigators. Confusion surrounded crucial tasks and frustration rose among people involved.
The cause of the explosion, which killed 11 workers, remains under investigation by federal authorities. But as more information trickles out, the image of a high-stakes, high seas venture collapsing in disarray is sharpening.
The mile-deep exploratory well was being capped with cement so it could be abandoned until a production phase later.
BP operated the well, Transocean owned the rig and Halliburton carried out the cement job. They had to work together.
Yet key plans kept changing. Critical tests meant to ensure the well would be safely cemented were not going smoothly.
Gagliano's computer model exposed yet another possible problem. The longtime technical adviser concluded that the cementing operation needed more centralizers, devices designed to ensure that the casing -- or drilling pipe -- runs down the center of the wellbore to increase the chance for a perfect seal and prevent leaks. BP had planned to run six centralizers and had them onboard.
After a chat with BP's senior drilling engineer, Gagliano worked up more models. By the evening of April 15, Gagliano had a model with 21 centralizers that resolved the gas flow problem. The 15 additional centralizers were acquired and scheduled for delivery the next morning. Still, the debate didn't end.
BP drilling engineer Brett Cocales learned the next afternoon, April 16, that his company's engineers had decided against using the additional centralizers because of questions about their mechanical integrity. Members of a BP investigation panel said Wednesday those concerns were unfounded because engineers were mistaken about which centralizers had been shipped.
Meanwhile, the well-lining job, known as casing, was completed. The casing, critical to ensuring no gas slips out of the wellbore uncontrolled, looked good. At 9 p.m. April 19, the Halliburton workers began the cement job 5,000 feet below the surface.
Sometime during the night, Nathaniel Chaisson, a Halliburton cementer, noticed a BP drilling engineer calling Houston headquarters. "We may have blown something out higher in the casing," the engineer said, triggering a flurry of phone calls between the rig and Houston.
Ultimately, they continued the job and Chaisson did not indicate there was ever discussion to halt the work.
Since then, however, BP has said it found weaknesses in the design and testing of the cement. It said the weaknesses may have allowed hydrocarbons to escape from the wellbore, causing the blowout. Halliburton insists the problem was not in the cement but in BP's design of the well.
By 1 a.m., on April 20, the cement job was completed.
Confidence was up. Yet at the 11 a.m. meeting, rig leader Jimmy Harrell was unhappy.
BP's plan for shutting down operations didn't include a "negative pressure test" -- a procedure that reduces the fluid pressure in the well to ensure there are no gas leaks. Harrell learned a long time ago this was risky. He demanded the test.
Meanwhile, in Houma, La., a group of VIPs -- BP executives Pat O'Bryan and David Sims, and Transocean executives Daun Winslow and Buddy Trahan -- boarded a helicopter headed for the Deepwater Horizon. They were going for a 24-hour "visibility" visit to one of their best-performing Gulf rigs.
They arrived on the rig floor as the negative test was going on. Winslow realized there was confusion over the pressures -- indicating there might be a leak somewhere in the well bore -- and decided it was not a good environment for the group.
Harrell and a Transocean toolpusher, Randy Ezell, stayed to help with the test. Bob Kaluza, a BP company man, was not satisfied. He got permission from Houston to run a second test.
An argument erupted over how to do the second negative test. Transocean toolpusher Jason Anderson wanted it done as it had always been done on the Horizon, and finally persuaded the BP crew. By 8 p.m., the test had ended.
"Go call the office," BP's well site leader Don Vidrine told his counterpart, Kaluza. "Tell them we're going to displace the well" -- a critical task in which the mud that keeps the oil and gas under control is replaced with seawater before the well is closed and abandoned.
Vidrine's decision to proceed may have been a fateful one. BP now thinks that Transocean and BP supervisors misinterpreted the results of the negative-pressure tests.
By 10 p.m. the first of two explosions shook the vessel.