Failing People Failed Machines: Judge’s ruling gives picture as to what caused BP environmental disaster

Plaintiffs eye dollars from $1.1B Halliburton settlement
September 25, 2014
A judge judges judging
September 25, 2014
Plaintiffs eye dollars from $1.1B Halliburton settlement
September 25, 2014
A judge judges judging
September 25, 2014

EDITOR’S NOTE: This story is solely sourced – except where otherwise indicated – from the findings of fact published by Judge Carl Barbier, of the U.S. District Court for the Eastern District of Louisiana, for the case titled “In re: Oil Spill by the Oil Rig ‘Deepwater Horizon’ in the Gulf of Mexico, on April 20, 2010.”

For purposes of clarity, brevity and accuracy, words and phrases from the judge’s own narrative are used extensively in this story without further attribution. Therefore, details presented as “fact” are considered so because the judge accepted them as such. During the translation process, oilfield experts were consulted from time to time to ensure factual quality. It should also be noted that many issues covered in the decision, especially those assigning blame or related to applicable laws, have not been addressed here.

On the evening of April 20, 2010, a blowout, explosions and fire occurred aboard the mobile offshore drilling unit Deepwater Horizon as it was in the process of temporarily abandoning a well, known as Macondo, it had drilled on the Outer Continental Shelf off the coast of Louisiana.

So begins the story – with some minor editing for style – of what may be one of the most significant court rulings of this century, which found that the international oil giant British Petroleum evinced gross negligence in connection with its operation of the rig during the abandonment process.

U.S. District Judge Carl J. Barbier’s 153-page ruling is certainly noteworthy for its result, which is still being litigated in the federal courts. But it also provides, for the first time, a cohesive, authoritative, detailed and objective account of that tragic night’s events and what led up to them.

Reading like a cross between a Sebastian Junger tell-all and a technical manual, Barbier’s decision provides intricate insight into how a series of errors and omissions – each independently harmless perhaps – conspired to produce a deadly and environmentally horrific accident.


Eleven men were killed as a result of the incident: Jason Anderson, Dewey Revette, Aaron “Dale” Burkeen, Donald Clark, Stephen Curtis, Roy “Wyatt” Kemp, Karl Kleppinger, Shane Roshto, Adam Weise, Keith Blair Manuel and Gordon Jones. At least 17 others were injured.

In the midst of frantic search and rescue efforts, the Deepwater Horizon burned continuously until mid-morning on April 22, when it capsized and sank into the Gulf of Mexico. As the rig descended, the marine riser — approximately 5,000 feet of pipe that connected the rig to its sub-surface blowout preventer — collapsed and broke.

Millions of gallons of oil discharged into the Gulf of Mexico over the next 87 days, causing environmental damage still undetermined, shutting down major fishing grounds in the Gulf and resulting in a flood of litigation that remains unsettled.

More than 3,000 cases with over 100,000 claimants’ alleged wrongful death and injury due to the explosion, later injury from exposure to oil and dispersants, damage to property or natural resources and economic losses resulting from the oil spill.

Businesses along the Gulf of Mexico coast that depended on deep-water drilling went under or survived barely due to a moratorium on exploration imposed by federal authorities convinced that a new disaster might occur.

“It just devastated the seafood industry to start off with and devastated our coast for the long run and we are still not sure,” remarked Terrebonne Parish Councilman Dirk Guidry, whose district includes many coastal communities. “It doesn’t seem like the shrimp are there and the oysters and the crabs. We are losing more land than we used to. It was horrific, and it was man-made. It wasn’t mechanical. It was more about money than safety.”


That profit trumped safety during BP’s operation was suspected from the beginning. But Barbier’s decision gives opinions like Guidry’s the stamp of judicial approval.

BP has voiced exception to Barbier’s findings on multiple counts, with good reason to do so. If they withstand scrutiny from the 5th Circuit U.S. Court of Appeals, to which BP has filed its objections, the company could eventually be liable for $21 billion in federal fines, not counting what has been paid out and may still be paid out in individual claims.

“BP believes that the finding that it was grossly negligent with respect to the accident and that its activities at the Macondo well amounted to willful misconduct is not supported by the evidence at trial,” says a statement from the company. “The law is clear that proving gross negligence is a very high bar that was not met in this case. BP believes that an impartial view of the record does not support the erroneous conclusion reached by the District Court.”

In addition to BP, Halliburton and Transocean were deemed negligent but not to the same extent, and not to a level reaching the recklessness that Barbier assigned to BP.

On Sept. 2, Halliburton settled with its claimants, agreeing to pay $1.1 billion.

Barbier found in his Sept. 4 ruling that BP bears a 67 percent responsibility, Transocean 30 percent and Halliburton 3 percent.


The Deepwater Horizon’s story begins in 1998, when a group of exploration and rig companies entered into a contract for construction of a semi-submersible mobile drilling rig capable of working in deep water.

Completed in 2001, the Horizon was 396 feet long and 256 across, floating on two massive pontoons. It was owned by Transocean Ltd., the world’s largest drilling operator.

Four large columns rose from the pontoon and supported the main deck, drill floor, derrick, bridge, engines and living quarters, as well as a helipad and cranes. The rig was a marvel, capable of drilling to a total depth of 35,000 feet in waters up to 10,000 feet deep.

It was classified as “semi-submersible” because it could partially submerge itself during operations, increasing stability. The rig was also self-propelled, meaning it could move about on a job site, powered by thrusters. They were also used while on assignments to hold the rig in place, rather than anchors or other, less stable attachments. At least one thruster was therefore active at all times.


On March 19, 2008, BP Exploration and Production Inc. acquired a lease from the United States of America, for 5,760 acres of property on the Outer Continental Shelf called the Mississippi Canyon Block 252, shortened in court documents to “MC252,” about 50 miles off the Louisiana coast. Initial plans were for the well to be drilled a total of 20,200 feet, measured from the rig deck. Transocean contracted to do the job, choosing a rig called the Marianas.

In simple terms, drilling the Macondo well involved a repeated sequence of steps: drill a certain distance, stop drilling, set “casing” to reinforce the wellbore, cement the casing in place, drill further, stop and set more casing.

Casing is essentially large diameter pipe that is placed inside a drilled-out section of the well. Once properly cemented in place, casing isolates the adjacent rock from the well.

The top of each successive casing string is set inside the previous one. Thus, the diameter of the well decreases slightly with each new piece of casing.

Setting casing also takes time. Consequently, there are economic incentives for a well operator to set as few casing strings as possible.


The Marianas commenced drilling at the site on Oct. 6, 2009, and the job did not go smoothly, prompting workers to call Macondo the “well from hell.”

One problem was related to the well’s geology. Crews encountered increasingly fragile sandstone below the surface, increasing the potential for difficulty.

As a well is drilled, it encounters different layers of rock, some of which contain fluids such as hydrocarbons or brine, within the rock’s pores. These fluids are under pressure and, if the rock has sufficient permeability, will flow into an area of lower pressure. When formation fluids unintentionally flow into the well-bore, the space that is actually drilled, a “kick” occurs.

Unchecked, a kick can develop into a “blowout,” an uncontrolled flow of formation fluids into the well-bore and possibly to the surface. Kick events involving oil or gas are particularly dangerous given the flammable nature of the hydrocarbons.


To avoid kicks, pressure in the well-bore must be greater than the pressure from the pores of any exposed formations of rock, sand or other material.

This is typically achieved by pumping a dense drilling fluid, commonly known as “drilling mud” or simply “mud,” into the well. Professionals regulate its consistency and weight, ostensibly with an eye to safety.

Generally speaking, if the “mud weight” is greater than the pore pressure, fluids should not migrate into the well.

This weight of the drilling mud is key to a delicate balancing act. While it must be heavy in order to maintain the pressure gradient, it must not be so heavy that it can fracture rock.

If rock is fractured, mud may escape into the formation, a concept called “lost returns.”

That can lower the pressure inside the well to a point where it becomes “underbalanced,” resulting in a kick. The same condition can also result in a subsea blowout inside the rock or other formation.

Federal rules regulate the balance and therefore the risks operators may take in achieving it.


On Oct. 26, 2009, with the drill at a depth of 8,970 feet, the Marianas experienced a “kick.” The well was shut in as corrective measures were taken, although BP opted to drill another 100 feet so that casing – the piping that lines the well – could be set in more stable shale rather than in the quirky sandstone. BP acknowledged that its decision to drill the additional 100 feet under certain conditions that were being experienced came with a risk of running into future problems, potentially creating an “uncontrollable well control event.”

As workers dealt with their problems related to the earth’s crust, new danger loomed for the Marianas.

A hurricane named Ida had formed in the Caribbean, first striking Nicaragua and then hitting the Yucatan Peninsula with winds of 105 mph. Although weakened to ex-tropical status, Ida still had enough punch while traveling the northern Gulf to damage the Marianas, and so the rig was relieved after drilling about 9,000 feet overall.

The Deepwater Horizon was at another well called Kodiak, and it was chosen to take on the Macondo job. It arrived onsite Jan. 31, 2010, and within days had its blowout preventer properly latched to the wellhead.

Among those on board were a marine crew, consisting of a master and chief mate, dynamic positioning operators, bosuns, able-bodied seamen, and ordinary seamen.

A drill crew was on board as well, consisting of tool-pushers, drillers, roustabouts and other workers.

Drilling resumed on Feb. 11, 2010.


Generally, all went well until March 8, 2010.

With the well drilled to a depth 13,250 feet in a higher-than-anticipated area of pore pressure, there was a “kick.” The well was shut in, but the formation collapsed around an area of pipe.

The pipe was severed with explosives and a cement plug was placed around the severed piece of pipe at the bottom of the well. The well was then sidetracked, meaning the DWH crew drilled around that pipe and then continued downward.

The cause of the March 8 kick was BP’s decision to drill faster than its geologists could analyze the data from the well, BP’s decision to ignore the information it did have, or both.

There additional problems, including “loss of total returns” at a depth of 18,260 feet, resulting in another shut-in. Losing total returns means that all of the mud pumped from the rig escaped through open fractures in the formation at the bottom of the well.


BP decided to spend the next five days pumping well-bore strengthening treatments in an effort to repair the formation and the mud-wright was reduced.

The formation remained in a “very fragile state” even though the well stopped losing returns.

The crew, at that depth, had encountered “primary reservoir sands,” meaning they had struck oil.

On April 9, BP opted to drill an additional 100 feet to ensure that the well was as far as it needed to be.

But doing so, as one witness noted, would mean working with almost no drilling margin, the safety factor that allows for effective balancing of pressures with drilling mud.

Petroleum geophysicist Alan Huffman, during the trial, called that decision “one of the most dangerous things I had ever seen in my 20 years’ experience” and “totally unsafe.”

After drilling the extra 100 feet BP called total depth.

At 18,360 feet the well had reached its primary objective sands, but not the deeper, secondary objectives. Although the planned total depth was to have been 20,200 feet, BP called it a wrap.

BP’s Geological Operations Coordinator stated, “Drilling ahead any further would unnecessarily jeopardize the well bore… At this point it became a well integrity and safety issue.”


At that point BP was $60 million, or 60-70 percent over budget and 54 days behind schedule. For each additional day the DWH remained, BP would lose $1 million. Also, the DWH had a full dance card.

The rig was scheduled to get to another well called Nile, and then on to the Kaskida well, where operations had to begin by May 16 or the lease would be lost.

After calling total depth, BP planned to set production casing inside the well and then temporarily abandon, meaning the well would be secured so that the operator could return at a later time to commence completion.

That would have involved placing cement or mechanical barriers in the well to replace the drilling mud and the blowout preventer.

Another rig would come at another time, drill through the temporary barriers and turn Macondo into a producing well.

The temporary abandonment procedure ultimately selected for the Macondo well involved a series of steps, which included setting a cement plug, known as “production casing cement” at the well’s bottom, exchanging some of the mud in the well with seawater and performing a pressure test, replacing all of the mud above 8,367 feet to seawater, and placing a second cement plug between 8,067 feet and 8,367 feet.


The next step was setting the production casing, the tubing that is set inside the well to protect the sides from the oil that would gush to the surface during production.

Various types of casing are used in different configurations. BP chose a single-tube style. There was concern among engineers about how much pressure the casing planned for use in production might take.

In preparation for that task, there were e-mails between engineers and other employees regarding safety concerns.

On April 16, BP engineer Brian Morel e-mailed Halliburton employee Preeti Paikattu to ask at what point the production casing would buckle if it landed atop an obstruction. The reply was that the casing would buckle if 30,000 pounds of compressional force was applied, and that buckling would occur at a low point.

Setting the cement plug in the bottom of the well was a key part of the procedure for abandonment, and as experts testified, there would be little room for error.

The purpose of the cement is to achieve “zonal isolation,” to keep the petroleum product-bearing zones from the well and to prevent those hydrocarbons from migrating into and up the well.

“It was intended that the production casing cement, once pumped into place and sufficiently hardened, would take the place of the drilling mud as the primary barrier to hydrocarbon influx,” is one way it is explained in the ruling.

The cementing job did not accomplish its purpose.

The right type of cement must be placed in the right place in order to create a barrier. If either component fails, then the cement will not achieve zonal isolation.

The cement job at Macondo was designed to place “foamed” cement in the narrow space between the production casing and the formation, an area known as the “annulus.”

To do this, cement on the rig had to be pumped down the production casing and then up the annulus.


Before cement could be pumped, however, a mechanical device known as a “float collar” had to be converted from a two-way valve to a one-way valve. A Weatherford M45AP float collar was located at a point 189 feet above the very bottom of the well, an area known as the “shoe track.”

When the float collar is converted to a one-way valve, fluid from above can flow down through the float collar.

BP was responsible for determining whether that critical conversion took place. If it doesn’t happen, the cementing should not begin.

On April 19 at around 2:30 p.m., after production casing was run into the well, the operation to convert the float collar began, after the casing was run down the hole. BP’s own “best practices” call for converting the float collar before running the casing.


BP was concerned that running the production casing with the float collar converted might create high surge pressures, which could further damage the fragile rock formation. Consequently, BP decided not to convert the float collar until after the production casing was fully run.

While sound reasons were given for that decision by BP, it nonetheless increased the risk that debris from the well would flow into and plug the float collar or another critical component, called a reamer shoe. A device called a “shoe filter” could have averted such a problem, but no filter was used.

When BP first attempted to convert the float collar, the rig crew could not circulate mud – used to determine suitability of the area for cementing – mud was pumped down the well but there was no return mud at the surface.

Pressure on the casing also started increasing during the attempted conversion. The only explanation given was that debris blocked the flow path, and it was likely that the path was blocked at two points or more.

Debris likely flowed up and around the auto-fill tube while the production casing was being run down the hole in unconverted mode.


When mud circulation could not be initially achieved, BP directed the rig crew to repeatedly increase and then bleed off the pressure in the well – a process BP called “rocking” – which was intended to clear the debris.

Nine attempts were made over the course of two hours to clear the blockage and convert the float collar. Each attempt, with the exception of the sixth, used greater pressure than the last. During the sixth attempt the pressure was kept the same as it was for the fifth attempt, but the pump rate was slightly increased. BP chose to keep pump rates low because it was concerned about damaging the fragile rock formation.

Circulation was achieved on the ninth attempt, at five times the pressure and less than one quarter the flow rate called for in Weatherford’s specifications for its device.

After circulation broke, the circulating pressure was significantly lower than predicted.

After noticing the rapid depressurization and/or the low circulating pressure, a Halliburton cement engineer, while standing on the rig floor, overheard BP Well Site Leader Bob Kaluza state “I need to make a phone call. We may have blown something higher up in the casing.”

Brian Morel, a BP drilling engineer who was on the rig at the time, wrote in an e-mail dated April 19 “Yah, we blew it at 3,142 (psi), but still not sure what we blew yet.”

BP chose to have Transocean change pumps, and improvement resulted. But it was still lower than expected.

It was BP’s conclusion on April 19 that the float collar converted, based on the fact that circulation was established, but never actually verified it.


Without verifying whether the float collar converted and without resolving whether something was “blown” or why the circulating pressure was low – other than to conclude, the next day, that the predicted pressures must have been incorrect – BP instructed Halliburton to commence the cement job. But the float collar never did convert.

There was also a failing in the casing, though nobody realized it at the time.

Ultimately, the conclusion on this portion of the operation was that cement was pumped through a breach and so was placed improperly.

As a result, hydrocarbons later entered the well casing.

Pumping of cement began at 7:30 p.m. on April 19, and ended at 12:30 a.m. the next day.

About 60 barrels, including 48 barrels of what is called “foamed” cement, were pumped altogether through the casing.

If the breach was greater than one square inch in diameter, virtually all of the cement would have passed through the rupture point. Hydrocarbon-bearing sands were thus left exposed to the well and had access into the casing.


Safety precautions including acoustic tests, to determine the integrity of the cement, were never used, including those that are in BP’s own best practices.

On the morning of April 20 a crew from Schlumberger, whose members could have performed a key test, was shipped back ashore.

There were problems with the quality and type of cement used as well, with BP opting to use cement from the prior Kodiak drilling operation, which was not suited to the Macondo task.

But ultimately, quality didn’t matter if the cement never reached its intended destination.


Abandoning operations continued, such as displacement of drilling mud by seawater, which began at 8 p.m. the night of April 20.

As seawater replaced the heavier mud, hydrostatic pressure within the well decreased. But at 8:52 p.m. the well allowed hydrocarbons to begin flowing into the bore.

Pumps were running at a constant rate, but between 9:01 and 9:08 p.m. drill pipe pressure increased by 100 psi.

But nobody noticed that something could be wrong.

An increase in pressure, however, is not necessarily indicative of a “kick.”

At 9:08 p.m. pumps were shut down so that a test could be conducted, to determine whether the incoming mud was free of hydrocarbons and therefore fit to be discharged into the ocean.

During the shutdown, the drill pipe pressure rose by 246 psi. It took longer than usual for fluid to stop flowing out of the well after the pumps were shut down. Again, none of those facts were noticed as signs of a kick.

There is some evidence that a flow check was conducted while the pumps were off, but no flow was seen coming from the well, an indication that the well was secure.


At around 9:14 p.m., Well Site Leader Don Vidrine ordered that the pumps be restarted and return fluids sent overboard.

At around 9:17 a large pressure spike registered on the rig’s kill line. Other pumps were shut down and personnel were dispatched to the pump room to investigate.

The spike was due to a blockage of unknown origin and operations resumed, except for the one that registered a problem.

At around 9:25 p.m., pressure slowly started to build on the kill line, increasing rapidly by 9:27 p.m. Somebody heard a Transocean toolpusher and driller talking about “differential pressure,” but there were no signs of great concern.

Pumps 3 and 4 were shut down by 9:30 p.m. Pump no. 1 was shut down by 9:31 p.m. With all pumps off, pressure started to build on the drill pipe.

Although circumstances should have dictated it, an immediate shut-in of the well did not occur.


Trying to further diagnose problems, a floor-hand opened a valve on the standpipe manifold around 9:36 p.m. The pressure on the drill pipe dropped significantly but not entirely. Pressure quickly built up to 1,400 psi upon closing the valve at 9:38 p.m. After the stand pipe was closed the drill crew began lining up the return fluids with the trip tank to perform a flow check.

At 9:40 p.m., the flow check was performed and significant flow from the well was noted. Mud began to spill onto the rig floor.

At 9:41 p.m., the Transocean drill crew activated the blowout preventer’s annular preventer, closing it to prevent any flow. But it did not fully seal around the drill pipe.

The preventer closed on the shoulder of a tool joint, the place where two segments of pipe meet.


Knowing there was a problem, the crew closed a diverter packer and routed the flow of hydrocarbons to the mud-gas separator, rather than away from the rig, which they could have done. It is possible that the default setting was to send the mud to the rig, to prevent pollutants from being discharged into the water, but there has never been a determination as to why the manual button, which could have been used for that purpose was not activated.

Quickly overwhelmed, the mud and gas separator sprayed water from its vents.

At 9:46 p.m., the DWH began to shake and suffered a jolt. A loud hissing noise began, as high-pressure gas reached the rig surface and began venting onto the DWH deck. The gas seeped into enclosed spaces as the crew worked to avoid disaster. The crew, meanwhile, attempted to activate their bore rams, which closed around the drill pipe around a minute later.

The master opened a door on the port side of the bridge and saw drilling mud raining down on the rig’s service vessel, and someone from the drill floor called the bridge and said, “we have a well control issue” or something similar, then hung up. The dynamic positioning operator, who answered the call attempted to call the drill floor back, but no one responded.

The number of activated gas alarms increased, and someone else called the bridge, said “well control situation” and hung up.


The Damon Bankston, the tender vessel, was notified of a problem and told to move away from the DWH by the Master.

The drill pipe pressure started rapidly increasing at around 9:47 p.m., indicating that the variable bore rams had sealed the annulus. This likely led the drill crew to believe they had regained control of the well. However, gas continued to vent onto the rig, and the gas alarms continued to sound.

At 9:49 p.m. the gas ignited and exploded, its ignition source still unknown, although multiple possibilities exist.

The tragedy of errors, omissions and failures continued, however.

An emergency device called the shearing blind ram, which is supposed to cut the pipe and seal the well, was not activated, although that could have been due to a power outage.

A battery backup system was likely not properly maintained and too old.

Seconds after the explosion an emergency system that could disconnect the rig from the well and allow it to float away was activated but to no avail. The hydraulic lines and other gear required for it to work was already destroyed.


The first Mayday call went out at 9:53 p.m., not because the captain ordered it but because a crew member took it upon herself to do so.

By that time the rig had sustained at least two explosions, lost power, and was afire.

Despite many of the failings cited on many counts, there were multiple instances of heroic acts.

Senior toolpusher Randy Ezell dug himself out of debris in the living quarters and found Wyman Wheeler and Buddy Trahan, both of whom were partially buried in debris and seriously injured, whom he uncovered.

Other crewmates soon arrived with a stretcher and evacuated Trahan, whose injuries were the most serious.

Ezell tried to lift Wheeler and walk him out by himself, but Wheeler was too injured to be moved without a stretcher. Wheeler told Ezell to leave him behind, but Ezell refused and instead waited with his crewmate for another stretcher to arrive, all while the rig burned about them.

Eventually another stretcher appeared and Wheeler was evacuated.


Four people jumped roughly 75 feet into the sea before the lifeboats were launched, and there was a shortage of them. Fully half could not be accessed due to the fire, but the overwhelming majority of evacuees were able to escape the rig via the two forward lifeboats. The record reflects the lifeboats waited as long as reasonably possible before deploying, which gave many the opportunity to board. The lifeboats deployed around 10:19 p.m. and 10:25 p.m. After the lifeboats launched, seven people evacuated in one of the three forward life rafts. These people were largely delayed because they were busy helping others evacuate.

The last life raft was launched by Captain Curt Kuchta and the senior dynamic positioning operator, Yancy Keplinger.

Consequently, these were the two last people on board the DWH, and they evacuated by jumping into the water. Once in the water, the life raft remained tied to the rig by a rope.

Some people jumped into the water to swim away from the intense heat. A fast rescue boat from the Bankston was nearby, but could not approach the raft due to fire on the water. Captain Kuchta swam to the fast rescue boat, retrieved a knife, swam back to the raft, and cut it free so the raft and those in it could be towed to safety.

Ezell and Wheeler, who was in a stretcher, evacuated in this raft.

Of the 126 souls aboard the DWH, 115 survived.

“It appears that everyone who survived the initial explosions managed to evacuate the rig,” the decision reads. “This is a testament to the safety training that Transocean did implement aboard its rig.

The crew, Judge Barbier wrote, “acted appropriately and bravely in the face of chaotic circumstances that are, frankly, difficult to genuinely understand.”

Responders battle the blaze after the Deepwater Horizon explosion. Problems with the rig began around 8 p.m. April 20. As seawater replaced the heavier mud, hydrostatic pressure within the well decreased. Pumps were running at a constant rate and drill pipe pressure increased, but no one noticed.