21/08/2013 The Path to Disaster The Deepwater Horizon BP’s disaster in the Gulf of Mexico Professor Patrick Hudson & Tim Hudson Hudson Global Consulting 4th Annual Plexus Industrial Safety Lecture Industrial Psychology Research Centre 14th August, 2013 Transocean Deepwater Horizon 1 21/08/2013 The accident • 20th April 2010 • Hydrocarbons escaped into the well while it was being underbalanced with seawater • The Blow Out Preventer (BOP) was activated late and failed to shut in the well • The riser was not disconnected • The escaping gas ignited • 11 dead • The drilling rig Deepwater Horizon sank after 2 days 2 21/08/2013 Conclusion • The accident was preventable • If BP’s Operating Management System OMS had been applied rigorously, the temporary abandonment could have been completed safely • The dominant failures were associated with no risk analyses or assessments despite major changes and problems with the well • The primary causes can be related back to BP’s organizational culture Mississippi Canyon 252 • Original plan to produce a well for an Anadarko facility • Poor offset data – Exploration well rather than a production well • Drilled by Transocean Marianas – The Pharos from Piper Alpha • Marianas forced off in late 2009 after hurricane damage • Replaced early 2010 by Deepwater Horizon • Described as “the well from hell” 3 21/08/2013 The well Drilling Margin • Three factors to consider • Pore pressure – Formation pressure acting in on the hole • Fracture gradient – Point at which rock breaks with mud pressure • Equivalent Circulating Density – Extra pressure due to mud pump pressure 4 21/08/2013 Drilling the well 5 21/08/2013 Well Design March 8th Dress Rehearsal • On March 8th there was a kick (influx of hydrocarbon) • The kick was detected late – There were crane operations at the time – The Sperry-Sun mud-logger was not believed • • • • • The well was shut in successfully Recovery required a sidetrack Costs including rig time exceeded $10M No MIA report was made to London The same crew was also on board on 20th April 6 21/08/2013 Temporary Abandonment April • Original plan to have a keeper well – to be completed for production at a later date • The well ran out of drilling margin at 18360 ft • Formation pressure 16,800 psi -1150 bar • They drilled on to get below the pay-zone • An initial plan was produced on April 12th • Feedback was that a pressure test for integrity was missing • The decision to set the lockdown sleeve was retained • The temporary cement plug was to be set in seawater Sequence of abandonment plans 7 21/08/2013 Procedure for temporary abandonment of the well Cementing the well • • • • • • • Run long string 7” casing Centralize production casing to avoid channelling Clean out the well – bottoms up Convert float collar Pump spacer Pump nitrogen foamed cement Test well integrity – Positive pressure test – Negative pressure test 8 21/08/2013 Centralizers • Centralizers are necessary to ensure the annulus is constant, otherwise cement leaves mud behind allowing the possibility of hydrocarbon escaping through the mud • Halliburton ran its Opticem software and recommended 21 centralizers • BP eventually ran with 6 without consulting Halliburton 9 21/08/2013 Centralizers on the rig, but not used Cleaning the well • A well needs to be cleaned by circulating mud • Circulation also breaks gel that sets if the mud is left alone • Ideal and standard is 1x or 1.5x bottoms up • Extra pumping adds pressure as an increase in ECD • BP rejected the Halliburton recommendation to run a full bottoms up 10 21/08/2013 Float Collar • A float collar is a two-way valve that can be converted to a one-way valve to ensure the cement does not flow back up the casing (Utubing) • The float collar has to be converted with a sufficient flow rate of mud • This was never achieved, but conversion was accepted because of the high pressure applied to the float collar (3142 psi) 11 21/08/2013 12 21/08/2013 Debris in the shoe Description of Event – Placement of Cement Choke Boost April 19th – 20th • 19:30 – 00:30 - Cement job pumped as planned. • 60 bbls cement • Foamed cement used to reduce risk of losses • 6 inline centralizers were spaced across the main pay • 00:35 – 7:00 – Seal assembly installed and pressure tested, but not locked down to wellhead. Proceeded to preparation for positive casing test. • ~7:30 – Decision made not to run Cement Bond Log (CBL) premised on minimal losses and lift pressures observed during cement displacement. Kill BOP Sea Floor Cement Mud Spacer Seawater 14.17ppg SOBM Influx Shoe – 17,168’ Cement TOC – 17,260’ FC – 18,115’ Top of Cement 17260’ 14.0ppg 13.1ppg 12.6ppg 12.6ppg Primary reservoir sands 12.6ppg Shoe – 18,304’ BP Confidential: Horizon Investigation Update 26 13 21/08/2013 Pressure Test • Positive pressure test to ascertain if the casing above the plugs will hold – Overbalance the well (i.e. pump up from above) • Negative pressure test to ascertain whether the cement barrier is effective – Underbalance the well by removing mud and replacing it with sea-water (pump out from above) – If the well is safe then there should be no change in pressure through the BOP – Proposal in MMS plan to measure on kill line rather than drill-pipe Description of Event - Positive Pressure Test 2700 PSI Choke Boost April 20th Kill Drill pipe Closed Blind Ram BOP BOP Sea Floor • 7:00 – 12:00 - Successful positive pressure test for production casing and seal assembly. –Low Pressure 250 psi –High Pressure 2700 psi • Note: The positive test is not designed to verify integrity of the shoe track (cement and float collar). Cement Mud Spacer Seawater 2700 psi Influx 14.17ppg SOBM Shoe – 17,168’ TOC – 17,260’ FC – 18,115’ 14.0ppg 13.1ppg 12.6ppg 12.6ppg Primary reservoir sands 12.6ppg Shoe – 18,304’ BP Confidential: Horizon Investigation Update 28 14 21/08/2013 Negative Pressure Test First Negative Pressure test 15 21/08/2013 Second Negative Pressure test Description of Event - Negative Pressure Test 1400 PSI Choke Boost 0 PSI Kill Closed Annular BOP Drill pipe BOP Sea Floor Viscous Spacer Cement Mud Seawater Spacer Seawater Influx 14.17ppg SOBM FC – 18,115’ • 15:04 – 15:56 – Displaced mud with sea water to underbalance the well. • 15:56 – 16:53 – Spacer inadvertently placed across the BOP due to leaking annular. • 16:53 – 17:52 – Test started by monitoring pressure on drill pipe. • 17:52 – 19:55 – Negative pressure test procedure switched to monitor pressure on kill line. – No pressure and no flow observed on the kill line – 1400 psi observed on the drill pipe • 19:55 – Shoe – 17,168’ TOC – 17,260’ April 20th • Negative test designed to verify integrity of seal assembly, casing, and shoe track in an underbalanced condition. This simulates conditions for T&A and rig demobilization. • 12:00 – 15:04 - Reviewed displacement procedure and commenced preparation for negative test. Test was concluded to be successful. 14.0ppg 13.1ppg 12.6ppg 12.6ppg Primary reservoir sands 12.6ppg Shoe – 18,304’ BP Confidential: Horizon Investigation Update 32 16 21/08/2013 Description of Event - Initial Recognition of Abnormal Well Conditions Choke Boost Kill Hydrocarbon Influx above BOP Seawater BOP Sea Floor April 20th Note: the following details are largely based on interpretation of data and witness statements: • 21:31 – After completely displacing the spacer out of the riser, the pumps were shut down. • 21:31 – 21:34 – Rig crew discussed abnormal drill pipe pressure increase. Mud • 21:36 – Rig crew bled the drill pipe to investigate the abnormal pressure. Spacer • 21:38 – Hydrocarbons began to enter riser. Cement Seawater • 21:40 – Mud overflowed the flow-line and onto rig. • 21:41 – Mud shot up through the derrick. Rig crew diverted to the mud gas separator (MGS) and shut the annular BOP. 21:41 Influx Shoe – 17,168’ TOC – 17,260’ 14.0ppg 13.1ppg 12.6ppg FC – 18,115’ 12.6ppg Primary reservoir sands 12.6ppg Shoe – 18,304’ BP Confidential: Horizon Investigation Update 33 Kick detection 17 21/08/2013 18 21/08/2013 12” Vent 6” Vacuum Breaker Bursting Disk M GS Rated to 60 psi Rotary Hose M ud System IBOP Hydrocarbons routed to MGS Port Overboard Starboard Overboard Diverter Flow Line Diverter Line Slip Joint Boost Diverter overboard lines closed MGS vented to manned and hazardous areas Starboard Overboard Rated to 100 or 500 psi Kill Choke BOP System quickly overwhelmed by pressure and volume, leading to loss of containment 19 21/08/2013 Blowout Preventer (BOP) Emergency Disconnect System (EDS) 20 21/08/2013 What was the immediate cause of the blowout? • Most common explanation in terms of failure of cement to provide an adequate barrier – But see an alternative explanation • With low expectations of a kick during abandonment (rather than during ordinary drilling) small and slow moving indicators were missed • Simultaneous operations for discharging together with major crane activities made observations hard – The Sperry-Sun mudloggers complained, but to no avail • Transocean’s emergency structure created problems with the use of the BOP and detachment from the riser Shoe Track & Cement 21 21/08/2013 How did this come to pass? BP 1990 - 2010 • BP was a medium-sized North Sea operator • 1990 John Browne appointed as CEO BP Exploration and Production (upstream) • 1995 Browne appointed CEO British Petroleum • Sequence of mergers, mostly in USA, to overtake Shell Group – Amoco – Arco – Burmah-Castrol • Number of disasters from 2000 onward – Grangemouth, Texas City, Thunder Horse, Prudhoe Bay – Market fixing prosecutions in USA 22 21/08/2013 Cost Cutting • Browne instituted a rigorous discipline of cost-cutting • This was rated as the cause of problems at both Texas City and Prudhoe Bay • Texas City had 2 25% across the board cost reduction targets prior to the disaster • The financial community loved Browne – The Sun King – for his financial discipline – even in an article published 20th April 2010 • Browne created a company that was loss averse, unlike its risk averse competitors (Shell and Exxon-Mobil) • Tony Hayward continued the culture with “Every dollar counts” 2000 Grangemouth • Number of major incidents, no fatalities • Pleaded guilty to criminal charges 23 21/08/2013 Texas City 2005 • BP’s Texas City refinery had a major explosion on March 23rd 2005 of the isomerization plant • 15 dead, 170 severely injured, >500 wounded • More than $ 1500 M set aside for compensation and $ 1000 M for remediation & improvement • Not including lost production 24 21/08/2013 Thunder Horse 2005 • July 2005, after hurricane Dennis in the GoM, BP personnel returned to find the platform with a 20o list • Start production in 2008, instead of original 2006 start-up • Losses (cost and 2 years lost revenue) exceeded $10 Billion Thunder Horse Platform, Gulf of Mexico BP Alaska Prudhoe Bay 2006 • 2006 a leak of crude (1m litres) from the North Slope to Valdez pipeline led to shutting down part of the pipeline • A second pipeline problem emerged leading to a major shutdown of production • Production losses were 400,000 bbl/day – Total 50,000,000 bbl ($3 Billion) • Reputation damage increased with 2nd closure • Scrutiny from US Congress • Fatality Nov 13 25 21/08/2013 Issues • Personal vs Process safety – Stop the Job – Workforce did not see process safety issues as relevant (more about environment than safety) • The contracts and safety responsibility – Bridging documents – Exclusion clauses • Foamed cement slurry stability • 1st and 2nd line kick detection Hypothesis – they blew the bottom off it • Proposal – backed up by logging from relief wells – Discussed by Chief Counsel but rejected for no reason • The bottom part of the casing , shoe track, was buckled or sheared • Transient block load of 140,000 lb (reported as 10,000) on final running of production casing – Evidence from Sperry-Sun recording of block load – This is ± 70 tonnes (about 3 double-decker busses) • Consequence that the cement never got near the payzone – Made blowout inevitable if the float collar failed – We know the annulus cement worked 26 21/08/2013 How did this come to pass? • • • • • Changes in well design Abandonment plan Lack of management of change Lack of formal risk management Lack of requirements for risk management 27 21/08/2013 Risk Assessments • The risk of catastrophic loss of containment was recognized as #1 at the Group level (London) and SPU level (Houston) • This was driven partly by downstream major incidents (Texas City, Grangemouth, Prudhoe Bay) • Macondo was downgraded from high to moderate • Kicks were downgraded from catastrophic loss of containment as consequence BP Group Risk Register 28 21/08/2013 No Drilling Surprises 29 21/08/2013 Design • Beyond the Best, Drilling and Well Operations Plan (DWOP) systems in place in GoM Drilling and Completions • gHSSEr old SMS was replaced after Texas City by OMS (Operations Management System) OMS • OMS was designed after Texas City as the solution to BP’s Process Safety management problems • Design started 2005, completed 2007 • Rollout late 2009 with USA as first wave • Local OMS (LOMS) to be defined for specific operations • BP assets (Texas City, Prudhoe Bay, Thunder Horse) rolled out first • Later rollout for contractor owned assets starting 2010 30 21/08/2013 The grandfather decision • Because MC-252 was already under construction and was planned to be finished by the time the Local OMS was rolled out BP decided not to implement the LOMS on the well and thus not subject it to the risk management processes it mandated. • “We decided not to grandfather it in” Risk decisions • Numerous risk decisions were made in the design of the well • Variances from technical standards were approved by the internal technical authority • These decisions were made by technically competent engineers in the engineering department and were subject to review 31 21/08/2013 Additional risk decisions • The temporary abandonment plan was not subjected to the same level of risk assessment and not reviewed at all • The regulator (MMS) approved the plan in less than 90 minutes • There were several changes to the well design that weren’t subjected to formal management of change nor any form of formal risk management • The plan implemented (Ops plan) deviated from the MMS approved plan Risk management • There was no process in place or in operation requiring and driving formal risk management of decisions • There was no management oversight of the decisions being made by the well team • The members of the well team did not have the skills to use the risk management tools provided by company • Silos between operations and engineering meant that each was waiting for the other to initiate risk assessments 32 21/08/2013 D&C senior engineer Q. Okay. All right. When you made the decision not to use the additional centralizers and to reposition this six existing centralizers — the subs, I believe, that you had on the rig — did you do a written risk assessment? A. No, ma'am. Q. Did you instruct anybody to do a written risk assessment? A. No, ma'am. Q. Was the risk register updated in connection with this decision? A. Not to my knowledge. Q. All right. And that was ultimately your responsibility, wasn't it? A. Yes, ma'am. D&C Well Team Leader Q. Do you know what the BP risk assessment tool is, the RAT? A. Yes. Q. Do you know how to use it? A. No. 33 21/08/2013 Drivers • Press-on-itis • Infrequent post drilling kicks • Focus on the next jobs (Nile P&A, Kaskida) Not weak signals • Throughout the abandonment process there were a number of signals and inputs that provided BP with information that there decisions were increasing the risk of failure of the operation • These were disregarded for a number of reasons 34 21/08/2013 Example Signals • The failure to catch the March 8th kick in a timely fashion was a signal that the well control process was not as robust as expected • The communication by the cementers that the decision not to carry out a full bottoms-up and the decision not to run all the centralizers placed the quality of the cement job in jeopardy Why were they ignored? • Lack of formal risk management processes meant that there was an incomplete understanding of the potential downsides of the decisions made • They had a purely forward looking point of view that failed to take account of their previous decisions that increased the overall risk 35 21/08/2013 FIDO • “I’m afraid we’ve blown something higher up in the casing string.” (Well Site Leader) • In the words of the Chief Counsel’s report, “the rig crew proceeded onward”. Did they take safety seriously? • About 100 STOP cards were issued every day • March 29th A dropped winch handle was found on the rig floor – 2 hour shutdown, team flown out specially • A senior management visit to the rig was on the rig when it blew out – Including BP and Transocean VPs • The problem was safety was seen as personal, rather than personal + process safety – A fire on March 28th had no equivalent response – BP used the word Integrity rather than safety for processes 36 21/08/2013 Production Safety Management System (SMS) Better defenses converted to increased production Protection 37 21/08/2013 Safety Management System (SMS) Production Best practice operations under SMS Protection Bly Report Swiss Cheese Picture 38 21/08/2013 The Court Case - MDL 2179 Magistrate Judge Sally Sushan Federal Judge Carl J. Barbier MDL 2179 • A large number of parties in civil litigation – – – – – – – – Plaintiffs (300,000 + in Plaintiffs’ Steering Committee) US Department of Justice Louisiana, Alabama (Florida & Mississipi later) BP Transocean Halliburton MI-Swaco Cameron Iron works • Separate from Criminal prosecutions – BP & Transocean pleaded guilty (only prosecutions) 39 21/08/2013 Plexus Industrial Safety Lecture Series Hosted by the Industrial Psychology Research Centre University of Aberdeen Details of previous Plexus lectures: 2012 Professor Stanton ‘Forensic Human Factors: Ladbroke Rail Crash’ 2011 Dr Mica Endsley ‘Situation Awareness: Research and Design’ 2010 Professor Erik Hollnagel ‘Resilience Engineering’ can be found on: www.abdn.ac.uk/iprc 40