A Sea in Flames

by Carl Safina

  Drilling fluid specialist Leo Lindner had spent four years on Deepwater Horizon.

  Q: “And what is a good negative test?”

  Lindner: “Where you don’t have any pressure up the kill. Of course … I haven’t been a witness to that many negative tests.”

  And that’s another thing: negative tests are not routine on exploratory wells. The Deepwater Horizon mainly drilled exploratory wells. This well was unusual, because it was an exploratory well that was being converted to a future production well that would later be reopened and tapped. Not all the crew were familiar with all these steps and procedures.

  Dr. John Smith: “Before they ever started the test, they’ve got enormously high pressure on the drill pipe.” That should have been, he noted, “a warning sign right off the bat.”

  Leo Lindner: “They decided to go ahead and try to do the first negative test. They bled off some pressure from the drill pipe and got fluid back. They attempted it again and got fluid back.”

  But as Dr. Smith had said: “If it’s a successful test, there’s no more fluid coming back.”

  The crew had been replacing heavier fluid with seawater. But Lindner was sufficiently worried by the initial results that at around 5:00 P.M., he ordered his coworker to stop pumping drilling fluid off the rig. He wanted to keep his foot on the emergency brake.

  At 5:30, Transocean’s subsea supervisor Chris Pleasant comes on duty in the drill shack. “My supervisor was explaining to me that they had just finished a negative test. Wyman Wheeler, which is the tool pusher, was convinced that something wasn’t right. Wyman worked to 6:00 P.M. By that time his relief come up, which is Jason Anderson, which is a tool pusher as well.”

  It was a bad time to change guards.

  But now it’s approximately ten minutes till 6:00. Bob Kaluza, the BP company man, tells Jason Anderson, “We’re at an all stop.”

  Kaluza’s relief, BP company man Don Vidrine, is scheduled to come on at 6:00 P.M.

  Chris Pleasant: “Jason Anderson, he’s convinced that it U-tubed. Where that U-tube’s at, I don’t know. But, you know, I guess we never really had a clear understanding. Anyway Jason is telling Bob that, ‘We want to do this negative test the way Ronnie Sepulvado does it.’ And Bob tells Jason, ‘No, we’re going to do it the way Don wants to do it.’ So, probably five minutes after 6:00 or something Don comes to the rig floor. Him and Bob talks back and forth for approximately a good hour.”

  They discuss possible causes for the fact that they’re reading pressure on the drill pipe but not on the kill line. Don Vidrine believes that if the pressure in the drill pipe was evidence of a surge of gas deep in the well, they would be seeing similar pressure in the kill line.

  Later question: “Based on industry standard ways of reading negative tests, you’re looking for something pretty simple, right? A zero on the drill pipe and a zero on the kill line; right?”

  Dr. John Smith: “Right.”

  Q: “And if you don’t see that, you need to be very concerned; right?”

  Smith: “Yes.” Dr. Smith further says, “We know there’s all this heavy spacer mud stuff in the well below the blowout preventer. Likely that mixture is what’s going back up into the kill line, holding the pressure back.”

  Smith adds, “We’re doing a test with a line that’s got this dense stuff in it. So, the symptoms are a successful test, but the reality is—it’s not a test at all. My opinion.”

  In other words: the only reason they’ve got zero pressure showing on the line they’re relying on is that the thick spacer material has gotten in; the line is clogged.

  After thirty minutes of staring at zero pressure on the kill line, the team is convinced that they’ve completed a successful negative test. Never mind that the drill pipe has 1,400 psi on it. They’ve convinced themselves that this was due to something Jason Anderson was calling a “bladder effect.”

  BP well site leader trainee Lee Lambert was later examined on this point.

  Q: “What was Mr. Anderson saying about the bladder effect? Can you tell us?”

  Lambert: “That the mud in the riser would push on the annular and transmit pressure downhole, which would in turn be seen on your drill pipe.”

  Q: “Was Mr. Anderson explaining why they were seeing differential pressure on the drill pipe versus the kill line?”

  Lambert: “Yes.”

  Q: “Okay. And did anyone say anything or disagree with Mr. Anderson’s explanation?”

  Lambert: “I don’t recall anybody disagreeing or agreeing with his explanation. At the time it did make sense to me. My lack of experience—. After learning things after the incident, it did not make sense to me, because the kill line and the drill pipe are open up to the same annulus, so in theory should see the same pressure.”

  Q: “And since then have you had an opportunity to study this so-called ‘bladder effect’?”

  Lambert: “I have not found any studies on the bladder effect.”

  In September 2010, BP’s internal investigation concluded: “According to witness accounts, the toolpusher proposed that the pressure on the drill pipe was caused by a phenomenon referred to as ‘annular compression’ or ‘bladder effect.’ The toolpusher and driller stated that they had previously observed this phenomenon. After discussing this concept, the rig crew and the well site leaders accepted the explanation. The investigative team could find no evidence that this pressure effect exists.”

  After the negative pressure test, Vidrine tells Bob Kaluza, “Go call the office. Tell them we’re going to displace the well.” They’re about to remove their fluid and replace it with seawater. Poised on a mountaintop, over an oil volcano, they’re about to release the brake.

  They’re in a bit of a hurry. But what about the cement job; had it cured correctly already? The negative test helped convince them that it had. But that was only because the kill line was clogged and they chose to explain away the pressure they were seeing on the drill pipe.

  The industry standard for judging the success of cement work, to best try to ascertain whether the cement is bonding to everything properly, is called a “cement bond log test.” Halliburton, which did the cement job, will later tell a Senate committee that a cement bond log test is “the only test that can really determine the actual effectiveness of the bond between the cement sheets, the formation and the casing itself.”

  Of course, because Halliburton did the cement job, its people would like to blame BP for not using the definitive test. They don’t want anyone focusing on their cement itself.

  Using sonic tools, a cement bond log test makes 360-degree representations of the well and can show where the cement isn’t adhering fully to the casing and where there may be paths for gas or oil to get in. In reality, even a cement bond log test is not perfect. But it is the best test going.

  Perhaps the most skilled people to do a cement bond log test work for the rig-servicing company Schlumberger. They’re on the rig on the morning of April 20, ready to get to work.

  BP decides instead to just rely on the pressure tests and other indicators that say that all’s well with the well. BP tells the Schlumberger workers that their services won’t be needed after all, and arranges for them to leave.

  John Guide explains: “Everyone involved on the rig site was completely satisfied with the job. You had full returns running the casing, full returns cementing the casing. Saw lift pressure, bumped the plug, floats were holding. So really all the indicators you could possibly get. So it was outlined ahead of time in the decision tree that we would not run a bond long if we saw these indicators. So the decision was made to send the Schlumberger people home.”

  As the Schlumberger folks board a helicopter and lift off the rig, oil and gas are already trying to get into the well, pushing hard on the cement. At 11:00 A.M., as the helicopter flies out of sight, eleven men on the rig have eleven hours left to live.

  The main critical error was in not recognizing that the drill pipe pressure they were re
ading during the pressure test indicated that gas was already getting in—and, therefore, that the cement job had failed.

  Why did it fail? People will speculate for months. Some will suggest that the cement was not allowed to set adequately before BP began altering the well pressure during the positive and negative pressure tests. Others will see that as irrelevant. Even the time required for the cement to harden at the pressure and temperature deep in the well will be subject to controversy.

  Not until September and October did some of the most important pieces of this puzzle start to fit into a clearer glimpse of what happened.

  First, as promised, let’s revisit the centralizers. In late September 2010, when the relief well finally intersects the original well, it will find no oil outside the casing above the oil-bearing zone in the rock. This will confirm that the oil and gas flowed first out of the sand, then down more than 80 feet outside the casing, then into the well casing and up through 189 feet of “shoe track” cement within the casing. This entire 270-foot run—down outside, then up inside the casing—was supposed to be filled with cement. It’s astonishing that the cement in the casing failed. The inner cement was designed to be a solid seven-inch-diameter, 189-foot-long plug.

  Though Halliburton had recommended twenty-one centralizers to help ensure a good cement job, BP used only six. But the other fifteen would have been placed above the zone bearing the oil and gas. Above that hydrocarbon zone where the other fifteen centralizers would have gone, the engineers poured 791 feet of cement into the gap between the casing and the well wall. That upper cement, above the oil and gas, remained sound. Cement failed in and below the main hydrocarbon zone. The part of the cement job that failed was where the centralizers were, and in the reach below them, and inside the casing. The flow of gas and oil was not up outside the casing but out of the sand, then down to the bottom of the well, then up inside the casing—despite the cement there—and then out to the surface.

  This seems to acquit the centralizers. So was there something wrong with the cement formulation itself?

  In its September 8, 2010, investigative report, BP will blame the cement. They’ll offer several reasons why the cement could have failed: contamination with drilling or spacer fluids, contamination among the three cement parts, or “nitrogen breakout,” in which the nitrogen breaks out of the foam and forms big gas pockets.

  BP will have a third party make cement samples designed to resemble Halliburton’s and test them in a lab. (They could not get cement samples from Halliburton because those are impounded as legal evidence.)

  BP’s report will claim that Halliburton’s foamed cement would have broken down in the well. BP will say Halliburton used a mixture containing 55 to 60 percent nitrogen, and that lab results indicate that “it was not possible to generate a stable nitrified foam cement slurry with greater than 50% nitrogen by volume at the 1,000 psi injection pressure.” The investigation team will conclude that “the nitrified foam cement slurry used in the Macondo well probably experienced nitrogen breakout, nitrogen migration and incorrect cement density. This would explain the failure.… Nitrogen breakout and migration would have also contaminated the shoe track cement and may have caused the shoe track cement barrier to fail.”

  Of course, BP’s investigation is suspect of pro-BP bias. But even if the BP report is self-serving and finger-pointing, this we do know: the cement did fail.

  Just before Halloween 2010, the president’s Oil Spill Commission will make its own explosive announcement: Halliburton officials knew weeks before the fatal explosion that its cement formulation had failed multiple tests—but they used the cement anyway.

  On March 8, 2010, Halliburton e-mailed results of one failed test to BP, but sent only the numbers; there’s no evidence that Halliburton specified that the numbers indicated failed testing. BP had overlooked Transocean’s warning of “severe gas flow,” and might also have not understood it had been given information predicting that the cement would fail.

  Halliburton altered the testing parameters, but the cement failed several tests. Finally, just days before the blowout, one last test indicated that the cement would remain stable. But Halliburton may not even have received the results of that final test before pouring the cement on April 19. BP definitely did not get notified that one of the formulations tested successfully. In other words, when Halliburton pumped cement into the BP well, both companies apparently possessed lab results indicating that what they were doing was unsafe.

  After the blowout, Halliburton will refuse to give its exact cement formulation to BP for independent testing for its September report. But in coming weeks the president’s Oil Spill Commission’s chief counsel, Fred H. Bartlit Jr., will persuade Halliburton to hand over its cement formulation by reminding its officials that anything they withhold from federal investigators will enlarge the Justice Department’s billowing civil and criminal charges. He’ll then ask Chevron to create a batch and test the mixture under various conditions. In all nine tests, the cement formulation will prove unstable. The unstable cement solidifies into a firm indictment of Halliburton and BP liability.

  At the stupendous pressures acting upon the oil and gas in the surrounding strata, even small cracks in the cement are enough to allow the flow rates that would send 60,000 barrels of oil a day out of the well.

  Something called a “float collar” enters the discussion. These flapping one-way valves were situated far down the well casing. Rig workers had a hard time getting them to close. They may not have sealed properly, and the fact that the oil and gas shot up the well casing indicates that the float collar’s valves also failed.

  Having mistakenly concluded that no hydrocarbons are coming into the well, the workers declare success at 7:55 P.M. on April 20.

  At 8:02 they begin displacing all the remaining heavy fluids with seawater. This will take over an hour. They know they’re near completion when the spacer comes back up to the surface.

  Returning fluids are usually directed into “pits” on the rig. This time, when the spacer reaches the surface, the crew directs the material directly overboard. This was their way, remember, of avoiding the requirement to bring it ashore and dispose of it as hazardous waste. While the material is being dumped overboard, certain flow meters, or “mudloggers,” are bypassed. In fact, they may have also been bypassed for much of the day as valuable returning drilling fluids were directed onto a waiting ship, before the spacer was just dumped directly overboard.

  BP’s September investigation will conclude: “The investigation team did not find evidence that the pits were configured to allow monitoring while displacing the well to seawater. Furthermore, the investigation team did not find evidence that either the Transocean rig crew or the Sperry-Sun mudloggers monitored the pits from 13:28 hours (when the offloading to the supply vessel began) to 21:10 hours (when returns were routed overboard).”

  Consultant Dr. John Smith will later testify that bypassing the flow-out meter amounted to “eliminating all conventional well control monitoring methods. That’s essentially in direct violation of the Minerals Management Service rules.”

  There’s another major wrinkle. Typical spacers are 180 to 200 barrels in volume, an amount that can be pumped out in fifteen to twenty minutes. But because the crew was trying to get rid of all the unused kill-pill material and bypass the solid-waste requirements by using unwanted material, this spacer was over 400 barrels. That meant that the rig crew had to spend an extra fifteen to twenty minutes or so pumping it overboard.

  This extra fifteen minutes occurred between 9:15 and 9:30. Had any crewmates been monitoring the flow through the meters, they would have seen some very irregular pressure and flow readings. Those fifteen minutes, fifteen crucial minutes of not monitoring the volume of their fluids, ended at 9:30 P.M., when they so clearly should have realized they had a problem. Those fifteen minutes could have saved the rig.

  Halliburton’s cement. M-I Swaco’s spacer. Transocean and BP’s misinterpretation
of the negative pressure test. BP’s push to replace all the heavy fluid with seawater. An observation comes to me via this e-mail from a friend: “My ex-brother-in-law was up for the weekend. He was a mud engineer on rigs all over the world, offshore and on. He says there are no excuses, the company man’s supposed to be in charge of everything. One thing he was very insistent on is that there’s no such noun as ‘drill’ on the rig. You can have drill bit, drill string, drill pipe, but a drill is what you use to find the lifeboats.”

  Now the crew is bypassing their monitors as their excess spacer is being dumped overboard.

  They’d slowed the pumps at approximately 8:50 P.M. in anticipation of the returning spacer. Slowing meant they should have seen reduced flow coming out, but the flow out actually increased. This was another indication that pressurized oil and gas were entering the well.

  Starting at approximately 9:01 P.M., without a change in pump rate, the drill pipe pressure increased from 1,250 psi to 1,350 psi. Another indicator. The pressure should have decreased at this time, not increased, because they were replacing fluid weighing 14.17 pounds per gallon with 8.6-pounds-per-gallon seawater. This increase should have gotten the rig crew’s attention.

  Over the ten-minute period from 8:58 to 9:08, they gained 39 barrels of fluid, the result of upward pressure in the well.

  BP’s September report will note: “No apparent well control actions were taken until hydrocarbons were in the riser.” In other words, gas had already gotten past the blowout preventer and was rushing the final mile to the surface.

  Now the rig crew begins sending returning fluids into a mud-gas separator with limited capacity. They may have thought this was just a “kick,” a belch.

  In fact, an enormous volume of methane was streaming into the well, shooting upward from miles below, expanding as the surrounding pressure lessened, pushing out the fluid above it, gathering itself into an accelerating blowout. In an awful irony, this would have been the time to send all the returning material overboard.