A Sea in Flames

by Carl Safina

  So in various ways this was going to be a difficult cement job. As late as the afternoon of April 14, BP was still reconsidering the chosen long string casing design, with its heavier reliance on the integrity of the cement deep at the well bottom. And the porous surrounding rock was on everyone’s mind. Cement has to be pumped in under some degree of excess pressure in order to fully fill the gap and get a good bond to the rock and sand on one side and to the outside of the production casing on the other. You need enough pressure both to keep the hydrocarbons contained and to force the cement against the sides, but too much pressure will inject the cement into the sand, and you’ll lose it. The team spent days determining how to approach the cement job. BP engineer Mark Hafle testified to this: “We were concerned that the pore pressure and frac gradient was going to be a narrow window to execute that cement job. That’s why we spent five days.” BP’s Brian Morel apologized to a colleague for asking yet another question about the design in an April 14 e-mail that he ended with this resonant comment: “This has been a nightmare well.” Hafle added, “This has been a crazy well for sure.”

  When BP won the lease to this piece of seabed, it held an in-house contest to name it. The winner, “Macondo,” came from the mythical town hewn from a “paradise of dampness and silence” in Gabriel García Márquez’s novel One Hundred Years of Solitude. In the novel, Macondo is an accursed place, a metaphor for the fate awaiting those too arrogant to heed its warning signs. What had seemed a nice literary allusion now carries ominous portent.

  More complications. Part of Jesse Gagliano’s task was to model the cement’s likely performance in this well and design a procedure that would get the cement to the proper locations. On April 15, he discovered some problems. This space between the casing and the wall of this well was very narrow. And the previous experience with lost drilling fluid indicated soft walls, requiring a low cement-pumping rate. These conditions contributed to a model predicting that if the casing moved too close to one side of the well-bore wall, drilling fluid could get left behind, creating pockets or channels where the cement would not distribute uniformly. That is, it wouldn’t fill in all of the space it needed to fill.

  To prevent a casing from getting too close to one side of a well bore, drillers slip flexible metal spring devices called “centralizers” over the casing so that it will stay centered in the well bore. By keeping the casing centered, centralizers help achieve good, even, thorough cementing between the casing and the well’s geological wall. In this case, BP had six centralizers. That number concerned Gagliano. On April 15 Gagliano e-mailed BP saying he’d run different scenarios “to see if adding more centralizers will help us.”

  BP’s Brian Morel replied, “We have 6 centralizers.… It’s too late to get any more to the rig. Our only option is to rearrange placement of these centralizers.… Hopefully the pipe stays centralized due to gravity.”

  But Jesse Gagliano continued his calculations. He determined that twenty-one centralizers should create an acceptably safe cement flow.

  And it wasn’t really too late. On April 16, BP engineering team leader Gregg Walz e-mailed BP project manager John Guide, saying that he’d located fifteen more centralizers that could be flown to the rig in the morning with “no incremental cost” for transporting them. “There are differing opinions on the model accuracy,” he wrote to Guide, “but we need to honor the modeling.” He added, “I apologize if I have overstepped my bounds.”

  The centralizers made the helicopter trip to the rig.

  But Guide expressed dismay at these particular centralizers’ design, the addition of new pieces “as a last minute decision,” and the fact that it would take ten hours to install them. He wrote, “I do not like this,” adding that he was “very concerned about using them.”

  Walz backed off.

  Later that afternoon BP’s Brian Morel wrote to his colleague Brett Cocales, “I don’t understand Jesse’s centralizer requirements.”

  Cocales replied, “Even if the hole is perfectly straight, a straight piece of pipe in tension will not seek the perfect center of the hole unless it has something to centralize it.” And then he added this: “But who cares, it’s done, end of story, will probably be fine and we’ll get a good cement job.”

  That was on April 16. It seems to suggest a certain willingness to add risk.

  That’s not how BP’s managers saw it. Guide later testified: “It was a bigger risk to run the wrong centralizers than it was to believe in the model.”

  But months later in September, BP’s own internal investigation concluded, “The BP Macondo team erroneously believed that they had received the wrong centralizers.”

  In late July 2010, examiners from BP contractors Anadarko, Transocean, and Halliburton questioned Guide on his decisions.

  Q: “That left you several days to get whatever centralizers you felt might be needed.”

  Guide: “I didn’t feel they were needed.”

  Q: “So what you’re telling me is that there was just no discussion among you between you and Mr. Walz about just waiting for the right centralizers? None, zip, zero, true?”

  Guide: “That subject never came up.”

  Q: “You still had time between the 16th and the 20th—”

  Guide: “Well, we didn’t know if we could find them. That subject never came up.”

  Q: “Sir, can you tell us the number of times, that you have personal knowledge of, that BP did not follow the recommendations of Halliburton in connection with the cementing of any of its jobs, if any?

  Guide: “I don’t know of any.”

  Well, perhaps we know of one. Halliburton’s Gagliano accepted BP’s decision and, on April 17 and 18, developed the specific procedure for pumping the cement. Gagliano created and sent one final cementing model out to the team on the evening of April 18.

  The model would later cause a firestorm for a particular page that no one at BP seems to have looked at. That page said that using only six centralizers would likely cause channeling; it also noted: “Based on analysis of the above outlined well conditions, this well is considered to have a SEVERE gas flow problem.” But with twenty-one centralizers, it added, “this well is considered to have a MINOR gas flow problem.”

  This report was attached to an e-mail sent to Guide on April 18, but it went unopened because the casing with just the six centralizers was already down the hole. Although BP had had days to get the centralizers, it was now too late to read the e-mail predicting severe gas-flow problems. Guide later testified: “I never knew it was part of the report.”

  The cement job will fail. But a few months later, in September 2010, BP’s own investigation will conclude, “Although the decision not to use twenty-one centralizers increased the possibility of channeling above the main hydrocarbon zones, the decision likely did not contribute to the cement’s failure.”

  That’s BP’s executives exonerating themselves, so season it with a grain of salt. But numerous industry analysts think centralizers are not the smoking gun. We’ll get back to that question later, but for now, it’s important to understand the distances involved. The recommended twenty-one centralizers were meant to keep the bottom 900 feet of casing evenly centered in the well. If the workers had had all twenty-one, they would have put fifteen above the span containing the oil and gas, four in the zone that held the oil and gas, and two below that zone.

  BP placed the six centralizers so as to straddle and bisect the 175 vertical feet of oil and gas–bearing sands deep in the well, at depths of around 18,000 feet. They placed two centralizers above the oil and gas zone, two in the zone, and two below it.

  But even if centralizers won’t be the smoking gun, the e-mail exchanges over the centralizers convey the sense that the BP team isn’t treating this endeavor with the utmost care. When red flags go up, BP’s decision makers seem rushed, rather than thorough.

  BP e-mails suggest that its personnel believed that any problem with cement could be remediated with additional
cement. And actually, that’s often what’s done; well cement jobs sometimes do fail. The reason why they fail is seldom precisely ascertained. Usually the failure is not catastrophic and the fix is to pump more cement in, then test it again. For this reason, the industry has developed several ways of testing the soundness of cementing jobs.

  But detecting problems assumes, of course, that the cementing job will be properly tested.

  The crew did their cementing over a five-hour period, starting at 7:30 P.M. on April 19. When they finished, at around 12:30 A.M., the calendar had turned over to April 20.

  Testing. More complications: when wells lose drilling fluid—as this one did weeks earlier—one possible solution is to send down a special mixture of fluids to block the problem zones in the well bore. Think of the stuff made to spray into a flat tire to seal it enough to get you home. A batch of this mixture is called a “kill pill.” It is a thick, heavy compound (16 pounds per gallon, compared to 14.5 for drilling fluid and 8.6 for seawater).

  Two weeks before the accident, when the rig had its serious 3,000-barrel loss of drilling fluid, the fluid specialists made up a kill pill and pumped it down to the problem zone. It didn’t seem to work, so they mixed up another batch: 424 barrels of a combination of two materials. But just as they were preparing to send this second kill pill down the hole, the losses stopped.

  They now had a thick, unused 424-barrel kill pill sitting in an extra tank, taking up space on the rig. To dispose of it they had two options: take it onto shore and treat it as hazardous waste or use it in the drilling process. The second choice would allow them to skirt the land-based disposal process and dump the compound directly into the ocean.

  The drilling fluid specialists got the bright idea of using the unused kill-pill material in a “spacer.” A spacer is a distinct fluid placed in between two other fluids. When you’re pushing different fluids down a well, you’ll often decide to use a spacer between the different fluids—between displacement fluid and drilling fluid, for instance—so that they won’t mix and so you can keep track of where things are. A spacer also creates a marker in the drilling flow, which allows the rig team to watch the fluid returns, to ensure that flow in equals flow out.

  Because BP didn’t want to have to dispose of the thick kill-pill material, they mixed it with some other fluid to create a spacer. BP’s vice president for safety and operations, Mark Bly, later said that using such a mixture was “not an uncommon thing to do.” The rig’s drilling fluid specialist, Leo Lindner, put it differently, saying, “It’s not something that we’ve ever done before.” At a government hearing in August, BP manager David Sims was asked if he had ever used a similar mixture as a spacer. “No, I have not,” Sims said.

  Down the hatch it goes. Just like that.

  Q: “What if you hadn’t used it that way, what would the rig have had to do; hazardous waste disposal, right?”

  Lindner: “Yes.”

  Q: “When these pills are mixed, have you ever heard anybody characterize it as looking like snot?”

  Lindner: “It wasn’t quite snotty.”

  Q: “But it was close?”

  Lindner: “It was thick. It was thick, but it was still fluid.”

  Q: “So it was very viscous?”

  Lindner: “Yes.”

  Q: “And really the only reason for putting those two pills down there was just to get rid of them; is that your understanding?”

  Lindner: “To my knowledge—well, it filled a function that we needed a spacer.”

  Chief engineer Steve Bertone later recalled that after the explosion, “I looked down at the deck because it was very slick and I saw a substance that had a consistency of snot. I can remember thinking to myself, ‘Why is all this snot on the deck?’ ”

  Back on the rig, Transocean installation manager Jimmy Harrell outlines the well-closing procedure. BP’s company man (later testimony is conflicting as to whether Kaluza or Vidrine was speaking) suddenly perks up. Interrupts. Says, “Well, my process is different. And I think we’re gonna do it this way.” Chief mechanic Douglas Brown will later testify that BP’s company man said, “This is how it’s going to be,” leading to a verbal “skirmish” with Transocean’s Jimmy Harrell, who left the meeting grumbling, “I guess that’s what we have those pincers for” (referring to the blowout preventer). Harrell will later testify that he was alluding to his concerns about risks inherent in the cementing procedure, but would say, “I didn’t have no doubts about it.” He’ll claim he had no argument but that “there’s a big difference between an argument and a disagreement.” Chief electronics technician Mike Williams, seated beside BP’s company man, will later recall, “So there was sort of a chest-bumping kind of deal. The communication seemed to break down as to who was ultimately in charge.”

  This is certainly not the time for chest bumping or blurred authority. If you’re gonna release the parking brake, you’d better agree on who’s gonna be in the driver’s seat. And whoever grabs the wheel better know how to drive.

  High-risk pregnancy enters labor. To determine if the cement job has worked and the well is sealed, rig operators can choose from several tests. On the Deepwater Horizon the engineers decide to do two kinds of pressure tests. In a “positive pressure test,” they introduce pressure in the well; if it holds, it means nothing’s leaking out from the well into the rock. They do this test on the morning of April 20, between about 11:00 a.m. and noon, roughly eleven hours after the cement job ends. It goes well; it seems nothing’s leaking out.

  But the reason nothing’s leaking out may be that there’s pressure from oil and gas pushing to get in. So the engineers prepare to do a “negative pressure test.” A negative test is a way of seeing if pressure is building in the well, indicating that gas and oil are leaking in. That could mean the cement has failed.

  To do a negative test, they close the wellhead, then reduce the downward pressure on the well by replacing some heavy drilling fluid with lighter water. Then they look at pressure gauges. If the pressure increases, hydrocarbons are entering, exerting upward pressure from below. What they want to see is zero pressure.

  Until the negative pressure test is performed successfully, the rig crew won’t remove the balance of the heavy drilling mud that stoppers the well; that’s their foot on the brake.

  Between 3:00 P.M. and 5:00 P.M., about fifteen hours after the cement job was finished, they start reducing the pressure by inserting seawater into the miles-long circulating-fluid lines. To make sure the drilling fluid and the seawater don’t mix, they precede the seawater with a spacer. The spacer they use contains that extra kill-pill material, the “snot.” And though a typical amount of spacer is under 200 barrels, this time it’s over 400 barrels because, remember, they’re trying to get rid of that leftover stuff.

  There are various places all this fluid is getting to, because there are various lines and pipes going into and out of the blowout preventer. One such line is called the “kill line.” Another is the drill pipe.

  A little before 5:00 P.M., they work for a while to relieve any residual pressure and are looking for the fluid to stabilize at zero pressure, indicated by a reading of zero pounds per square inch, or psi. They’ve got the pressure down to 645 psi in the kill line, but it’s at 1,350 psi in the drill pipe. So they try bleeding the system down, venting off some of that pressure. They achieve zero in the kill line. The drill pipe retains 273 psi. They need zero.

  Over six minutes right around 5:00 P.M., the drill pipe pressure increases from 273 psi to 1,250.

  The engineers tighten the blowout preventer’s rubber gasket and add 50 barrels of heavier-than-water fluid.

  So the lines are filled with a variety of different fluids snaking through in segments: there’s a stretch of drilling fluid, or “mud,” a stretch of the unusual spacer material, followed by plain seawater. At this point in the circulation of the various fluids, the spacer—the “snot”—should be above the blowout preventer. But some of it has found its wa
y into the blowout preventer and has entered one of the lines being tested.

  The engineers see pressure building in the drill pipe, zero pressure in the kill line. They’re unsure what to make of that, so they repeat the test procedure several times. From shortly after 5:00 to almost 5:30, they get the pressure in the drill pipe down a little, from 1,250 to almost 1,200.

  The Deepwater Joint Investigation panel asked Dr. John R. Smith, whose PhD is in petroleum engineering, to describe a negative test:

  “If it’s a successful test, there’s no more fluid coming back. You’ve got a closed container. There’s no hole in the boat. There’s no fluid leaking in through the wall of the container or the casing. It just sits there. If you have an unsuccessful test, external pressure is leaking through the wall of the system somewhere. Through the wall of the casing, past the casing hanger seals, up through the float equipment in the casing—. Somewhere there’s a leak from external pressure into the system. You’d expect to continue to see some fluid coming back.”

  To reach for an analogy: if you did a negative test in a swimming pool, you’d empty the pool. If the pool stayed dry, you’d have a successful test. If you had a problem, the pool would start filling itself through leaks in its walls. A well 18,360 feet deep is a bit trickier to test than a swimming pool. Even though you’re reducing the pressure, you still have to keep enough downward pressure on the well to control any oil and gas that might start entering. And you must check specific pipes for indications of pressure.

  Wells come in many different sizes and shapes and pipe setups. So, somewhat surprisingly, there isn’t a “standard” negative test.

  Q: Do you know if there’s any standard negative test procedure that the industry follows?

  Dr. John Smith: I was unable to find a standard.

  The Minerals Management Service’s permit specified that this negative test be conducted by monitoring the kill line above the blowout preventer. John Guide: “And that was really the only discussion, was to make sure that we did it on the kill line so that we would be in compliance with the permit.”