Skip to content

Williams Olefins Plant Explosion and Fire

Overview

On June 13, 2013, at the Williams Geismar Olefins Plant in Geismar, Louisiana, a reboiler rupture led to a fire and explosion. Two workers were fatally injured. 167 reported workers were injured. The incident involved the propylene fractionator reboiler system and released over 30,000 pounds of flammable hydrocarbons. The plant remained down for 18 months and restarted in January 2015.

Incident Snapshot

Field Value
Facility / Company Williams Olefins LLC
Location Geismar, LA
Incident Date 06/13/2013
Investigation Status The CSB's investigation was released at a news conference in Baton Rouge, LA, on 10.19.2016
Accident Type Chemical Manufacturing - Fire and Explosion
Final Report Release Date 10/19/2016

What Happened

  • On June 13, 2013, during a daily morning meeting with operations and maintenance personnel, the plant manager noted that the quench water flow through the operating propylene fractionator reboiler (Reboiler A) had dropped gradually over the past day.
  • The group then analyzed plant data and noticed the entire quench water circulation rate seemed to be impaired.
  • An operations supervisor informed the group that he would try to determine what caused the drop in flow.
  • After evaluating the quench water system in the field, the operations supervisor informed several other personnel that fouling within the operating reboiler (Reboiler A) could be the problem and they might need to switch the propylene fractionator reboilers to correct the quench water flow.
  • The operations supervisor attempted to meet with the operations manager to discuss switching the reboilers, but the operations manager was not available.
  • The operations supervisor decided to return to the field and continue evaluating the quench water system.
  • The CSB determined that at 8:33 am, the operations supervisor likely opened the quench water valves on the offline reboiler, Reboiler B, as indicated by the rapid increase in quench water flow rate.
  • Approximately three minutes later, Reboiler B exploded.
  • Propane and propylene process fluid erupted from the ruptured reboiler and from the propylene fractionator due to failed piping.
  • The process vapor ignited, creating a massive fireball.
  • The force of the explosion launched a portion of the propylene fractionator reboiler piping into a pipe rack approximately 30 feet overhead.
  • A Williams operator working near the propylene fractionator at the time of the explosion died at the scene.
  • The operations supervisor succumbed to severe burn injuries the next day.
  • The explosion and fire also injured Williams employees and contractors who were working on a Williams facility expansion project.
  • The fire lasted approximately 3.5 hours.
  • Williams reported releasing over 30,000 pounds of flammable hydrocarbons during the incident.
  • The plant remained down for 18 months and restarted in January 2015.

Facility and Process Context

  • The Williams Geismar Olefins Plant employs approximately 110 people.
  • The plant is located in Geismar, Louisiana, approximately 20 miles southeast of Baton Rouge.
  • The plant produces ethylene and propylene for the petrochemical industry.
  • At the time of the incident, Williams Olefins LLC and Saudi Basic Industries Corporation (SABIC) jointly owned the plant, and Williams Olefins was the sole operator.
  • At the time of the incident, approximately 800 contractors worked at the Williams Geismar facility on an expansion project.
  • Reboiler EA-425B was a six-pass, straight tube, divided flow shell design constructed in 1967.
  • The subject reboiler EA-425B was off-line and blocked in, all process stream and quench water valves closed.
  • Operators adjusted quench water flow in the system to compensate for fouling in EA-425A by opening the quench water inlet and outlet valves of the subject reboiler EA-425B, while the process inlet and outlet valves remained closed.
  • The parallel reboiler EA-425A was on-line.
  • The system had been in this configuration for 16 months.
  • The reboiler supplied heat to a propylene fractionator column.

Consequences

  • Fatalities: Two fatalities
  • Injuries: 167 reported injuries
  • Environmental release: Williams reported releasing over 30,000 pounds of flammable hydrocarbons during the incident.
  • Facility damage: The reboiler shell catastrophically ruptured. The force of the explosion launched a portion of the propylene fractionator reboiler piping into a pipe rack approximately 30 feet overhead. The blast effects flattened the reboiler shell.
  • Operational impact: The fire lasted approximately 3.5 hours. The plant remained down for 18 months and restarted in January 2015.

Key Findings

Immediate Causes

  • The reboiler shell catastrophically ruptured, causing a boiling liquid expanding vapor explosion (BLEVE) and fire.
  • The introduced heat increased the temperature of the liquid propane mixture confined within the reboiler shell, resulting in a dramatic pressure rise within the vessel due to liquid thermal expansion.
  • The reboiler was isolated from its pressure relief device.
  • Fracture of reboiler EA-425B likely initiated in the outlet pad longitudinal seam weld and the growing crack initiated fracture in the shell heat affected zone of the outlet pad to shell fillet weld, at initiation site 2.

Contributing Factors

  • The incident occurred during nonroutine operational activities that introduced heat to a type of heat exchanger called a “reboiler” which was offline.
  • Process safety management program weaknesses at the Williams Geismar facility during the 12 years leading to the incident caused the reboiler to be unprotected from overpressure.
  • Deficiencies in implementing Management of Change (MOC), Pre-Startup Safety Review (PSSR), and Process Hazard Analysis (PHA) programs.
  • The company did not perform a hazard analysis or develop a procedure for the operational activities conducted on the day of the incident.
  • Williams did not perform adequate Management of Change (MOC) or Pre-Startup Safety Reviews (PSSRs) for two significant process changes involving the propylene fractionator reboilers—the installation of block valves and the addition of car seals.
  • Williams did not adequately implement action items developed during Process Hazard Analyses (PHAs) or recommendations from a contracted pressure relief system engineering analysis.
  • Williams did not perform a hazard analysis and develop a procedure prior to the operations activities conducted on the day of the incident.
  • Williams did not effectively use the hierarchy of controls in the 2001 design change that added block valves to the propylene fractionator reboilers.
  • Williams relied upon administrative controls to mitigate a serious overpressure hazard.
  • The 2006 PHA recommendation was marked “Complete” more than three years later in January 2010 in Williams’ action item tracking system, but was not implemented as the PHA team had intended.
  • The contracted PHA facilitator was under the incorrect impression that both propylene fractionator reboilers operated at the same time.
  • Williams did not perform a field verification of the documented safeguards as part of the 2011 PHA.
  • Williams did not develop an action item to address the relief valve engineering analysis for the propylene fractionator reboilers.
  • Williams did not effectively measure leading and lagging process safety indicators before the incident.
  • Chevron marks, indicating fracture directions, were present on the majority of the longitudinal fracture surfaces.
  • Macroscopically brittle fracture morphology with small shear lips predominated, indicating mode I loading.
  • There was a significant proportion of slant morphology indicative of mode III loading.
  • No evidence was found of cracks pre-existing the rupture event.
  • The chemical composition and tensile properties of the ruptured shell EA-425B and the parallel EA-425A reboiler shell met the requirements for ASTM A212 Grade B Firebox steel.
  • Charpy V-notch impact toughness testing indicated that the EA-425A shell met the requirements ASME Section VIII, Figure UG-84.1, at 50 °F and above.
  • The shell fracture morphology was cleavage.
  • Fracture toughness testing at 180 °F indicated good fracture toughness and stable crack growth up to nearly 7-mm (0.28-inches).
  • Evidence of ductile fracture was found in the outlet pad seam weld above fracture initiation site 2, and fracture model results indicated the outlet pad seam weld as a likely location of fracture initiation.
  • Assuming ductile fracture had initiated first in the outlet pad seam weld, the growing crack and resulting stress state just below the pad to shell fillet weld likely promoted cleavage fracture initiation at site 2.
  • Because on-site post rupture corrosion was most extensive at the north end of the reboiler where fracture initiated, some ductile character at initiation site 2 may have been obscured or destroyed by corrosion.
  • Fracture modeling indicated pressures required to initiate fracture were well above the maximum allowable working pressure and hydrostatic test pressure of the reboiler shell.

Organizational and Systemic Factors

  • Weaknesses in these programs resulted from a culture at the facility that did not foster and support strong process safety performance.
  • The company did not identify and control the new process overpressurization hazard.
  • The MOC was an after-the-fact activity for Williams to address a regulatory requirement rather than an effective tool used to identify and control new process hazards prior to installing the new equipment.
  • The MOC reviewers focused largely on managing documentation and maintenance requirements for the new valves, such as needed process safety information updates and inspection requirements, and not on how the addition of the valves could affect the operability and safety of the overall process.
  • Management’s approval of incomplete documentation can lead to a culture of complacency and, therefore, subpar and incomplete process safety analyses.
  • Williams did not have a policy requiring the effectiveness of safeguards to be analyzed.
  • Operations personnel had informal authorization to manipulate field equipment as part of assessing process deviations without first conducting a hazard evaluation and developing a procedure.
  • The deficiencies listed above highlight that both a strong written safety management system and effective implementation of that system are required to have good process safety performance.
  • In the years leading up to the incident, Williams Geismar exhibited characteristics of a weak process safety culture.
  • significant weaknesses in the Williams process safety culture
  • deficiencies in implementing the plant’s process safety management programs as well as weaknesses in the written programs themselves
  • poorly conducted Management of Change and Pre-Startup Safety Reviews
  • ineffective safeguard selections
  • insufficient safeguard evaluation requirements
  • poor implementation of action items in Process Hazard Analyses
  • inadequate focus on development and maintenance of operating procedures
  • uncontrolled field equipment manipulations without a hazards assessment prior to the development of a procedure

Failed Safeguards or Barrier Breakdowns

  • The reboiler lacked adequate overpressure protection.
  • The reboiler was isolated from its protective pressure relief valve located on top of the propylene fractionator.
  • The MOC reviewers did not identify that the reboilers required overpressure protection.
  • The MOC reviewers incorrectly indicated that existing operating procedures were adequate to account for the new valves.
  • The MOC reviewers improperly indicated that the change did not require a Process Hazard Analysis (PHA).
  • The MOC reviewers selected incorrect responses regarding whether the new equipment met all applicable codes and standards.
  • The PSSR question regarding whether a process hazard analysis had been completed, recommendations resolved, and incorporated in design as deemed appropriate was not answered.
  • The PSSR questions regarding operator training were not answered.
  • PSSR reviewers incorrectly answered “yes” to the questions “Are all necessary operating procedures in place and current for safety, environmental, operating, emergencies, maintenance and technical?” and “Are procedures available for new and modified equipment?”
  • No response was given to the question, “PRV’s [pressure relief valves] lined up and block valves car sealed open? Pressure release systems in place and operational and traced where appropriate?”
  • The 2001 PHA team did not identify reboiler overpressure as a possible safety consequence.
  • The 2006 PHA recommendation to lock open at least one of the manual valves associated with each of the propylene fractionator reboilers was not implemented as intended.
  • The shell-side valves of the standby reboiler remained closed, with no car seals on the manual valves and no protective pressure relief device installed on the shell.
  • Williams did not perform an MOC or a PSSR for the installation of the car seal.
  • Williams did not perform a field verification of the documented safeguards as part of the PHA.
  • The relief valve engineering analysis identified that the propylene fractionator reboilers did not have sufficient overpressure protection, but Williams did not develop an action item to address it.
  • The reboiler block valves were leak tested following the incident and their leakage rate was within that allowed by API Standard 598.
  • pressure relief device
  • administrative controls
  • Management of Change
  • Pre-Startup Safety Reviews
  • safeguard selections
  • safeguard evaluation requirements
  • action items in Process Hazard Analyses
  • operating procedures
  • hazards assessment

Recommendations

  1. 2013-03-I-LA-R1 | Recipient: Williams Geismar Olefins Facility | Status: Closed – Acceptable Action | Summary: Implement a continual improvement program to improve the process safety culture at the Williams Geismar Olefins Plant. Ensure oversight of this program by a committee of Williams personnel that, at a minimum, includes safety and health representative(s), Williams management representative(s), and operations and maintenance workforce representative(s). Ensure the continual improvement program contains Process Safety Culture Assessments and Workforce Involvement. Conduct the process safety culture assessments at least once every five years.
  2. 2013-03-I-LA-R2 | Recipient: Williams Geismar Olefins Facility | Status: Closed – Acceptable Action | Summary: Develop and implement a permanent process safety metrics program that tracks leading and lagging process safety indicators. Include components to measure the effectiveness of the Management of Change (MOC) program, the Pre-Startup Safety Review (PSSR) program, the methods to effectively and timely complete action items developed as a result of Process Hazard Analyses (PHAs), Management of Change (MOC), incident investigations, audits, and safety culture assessments, and the development and implementation of operating procedures.
  3. 2013-03-I-LA-R3 | Recipient: Williams Geismar Olefins Facility | Status: Closed – Acceptable Action | Summary: Develop and implement a program that demands robust and comprehensive assessments of the process safety programs at the Williams Geismar facility, at a minimum including Management of Change, Pre-Startup Safety Review, Process Hazard Analyses, and Operating Procedures. Engage an expert independent of the Geismar site to lead these assessments at least once every three years.
  4. 2013-3-I-LA-R4 | Recipient: American Petroleum Institute | Status: Closed – Acceptable Alternative Action | Summary: Strengthen API Standard 521, Pressure-relieving and Depressuring Systems, by defining the various types of equipment operating statuses. Include definitions for “standby” and “out-of-service.” Specify pressure relief requirements for each type of equipment operating status.
  5. 2013-3-I-LA-R5 | Recipient: American Petroleum Institute | Status: Closed – Acceptable Alternative Action | Summary: Clarify API Standard 521, Pressure-relieving and Depressuring Systems, to require a pressure relief device for overpressure scenarios where internal vessel pressure can exceed what is allowed by the design code.

Key Engineering Lessons

  • A reboiler or similar heat exchanger that can be isolated from its pressure relief device must not rely on administrative controls alone for overpressure protection.
  • Process changes such as adding block valves or car seals require effective Management of Change and Pre-Startup Safety Review to identify new overpressure hazards before equipment is placed in service.
  • Hazard analyses must explicitly consider the operating status of standby or offline equipment and verify that documented safeguards are actually present and effective in the field.
  • If a relief system engineering analysis identifies insufficient overpressure protection, it must be converted into an action item and tracked to completion.
  • Field manipulation of process equipment to address deviations should not occur without a hazard evaluation and a written procedure.
  • Process safety culture and metrics are necessary to detect whether MOC, PSSR, PHA, and operating procedure systems are functioning as intended.

Source Notes

  • Priority 1 final report was used to resolve conflicts and establish the authoritative incident sequence, causes, consequences, and recommendations.
  • Priority 4 supporting metallurgical analysis was used only for fracture initiation and material findings that did not conflict with the final report.
  • Priority 4 recommendation status pages were used for recommendation status updates and API closure actions.
  • Official terminology from the source documents was preserved where possible, including 'Management of Change (MOC)', 'Pre-Startup Safety Review (PSSR)', 'Process Hazard Analysis (PHA)', and 'Closed – Acceptable Action'.

Similar Incidents

Incidents sharing the same equipment, root causes, or hazard types.

Same Equipment

Same Root Cause

Same Hazard

← View in Knowledge Graph